Pipelines

EvalML pipelines.

Package Contents

Classes Summary

ARIMARegressor

Autoregressive Integrated Moving Average Model. The three parameters (p, d, q) are the AR order, the degree of differencing, and the MA order. More information here: https://www.statsmodels.org/devel/generated/statsmodels.tsa.arima_model.ARIMA.html.

BinaryClassificationPipeline

Pipeline subclass for all binary classification pipelines.

CatBoostClassifier

CatBoost Classifier, a classifier that uses gradient-boosting on decision trees. CatBoost is an open-source library and natively supports categorical features.

CatBoostRegressor

CatBoost Regressor, a regressor that uses gradient-boosting on decision trees. CatBoost is an open-source library and natively supports categorical features.

ClassificationPipeline

Pipeline subclass for all classification pipelines.

ComponentGraph

Component graph for a pipeline as a directed acyclic graph (DAG).

DecisionTreeClassifier

Decision Tree Classifier.

DecisionTreeRegressor

Decision Tree Regressor.

DFSTransformer

Featuretools DFS component that generates features for the input features.

ElasticNetClassifier

Elastic Net Classifier. Uses Logistic Regression with elasticnet penalty as the base estimator.

ElasticNetRegressor

Elastic Net Regressor.

Estimator

A component that fits and predicts given data.

ExtraTreesClassifier

Extra Trees Classifier.

ExtraTreesRegressor

Extra Trees Regressor.

FeatureSelector

Selects top features based on importance weights.

KNeighborsClassifier

K-Nearest Neighbors Classifier.

LightGBMClassifier

LightGBM Classifier.

LightGBMRegressor

LightGBM Regressor.

LinearRegressor

Linear Regressor.

LogisticRegressionClassifier

Logistic Regression Classifier.

MulticlassClassificationPipeline

Pipeline subclass for all multiclass classification pipelines.

OneHotEncoder

A transformer that encodes categorical features in a one-hot numeric array.

PerColumnImputer

Imputes missing data according to a specified imputation strategy per column.

PipelineBase

Machine learning pipeline.

ProphetRegressor

Prophet is a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. It works best with time series that have strong seasonal effects and several seasons of historical data. Prophet is robust to missing data and shifts in the trend, and typically handles outliers well.

RandomForestClassifier

Random Forest Classifier.

RandomForestRegressor

Random Forest Regressor.

RegressionPipeline

Pipeline subclass for all regression pipelines.

RFClassifierSelectFromModel

Selects top features based on importance weights using a Random Forest classifier.

RFRegressorSelectFromModel

Selects top features based on importance weights using a Random Forest regressor.

SimpleImputer

Imputes missing data according to a specified imputation strategy.

StackedEnsembleClassifier

Stacked Ensemble Classifier.

StackedEnsembleRegressor

Stacked Ensemble Regressor.

StandardScaler

A transformer that standardizes input features by removing the mean and scaling to unit variance.

SVMClassifier

Support Vector Machine Classifier.

SVMRegressor

Support Vector Machine Regressor.

TargetEncoder

A transformer that encodes categorical features into target encodings.

TimeSeriesBinaryClassificationPipeline

Pipeline base class for time series binary classification problems.

TimeSeriesClassificationPipeline

Pipeline base class for time series classification problems.

TimeSeriesFeaturizer

Transformer that delays input features and target variable for time series problems.

TimeSeriesMulticlassClassificationPipeline

Pipeline base class for time series multiclass classification problems.

TimeSeriesRegressionPipeline

Pipeline base class for time series regression problems.

Transformer

A component that may or may not need fitting that transforms data. These components are used before an estimator.

VowpalWabbitBinaryClassifier

Vowpal Wabbit Binary Classifier.

VowpalWabbitMulticlassClassifier

Vowpal Wabbit Multiclass Classifier.

VowpalWabbitRegressor

Vowpal Wabbit Regressor.

XGBoostClassifier

XGBoost Classifier.

XGBoostRegressor

XGBoost Regressor.

Contents

class evalml.pipelines.ARIMARegressor(time_index=None, trend=None, start_p=2, d=0, start_q=2, max_p=5, max_d=2, max_q=5, seasonal=True, n_jobs=- 1, random_seed=0, **kwargs)[source]

Autoregressive Integrated Moving Average Model. The three parameters (p, d, q) are the AR order, the degree of differencing, and the MA order. More information here: https://www.statsmodels.org/devel/generated/statsmodels.tsa.arima_model.ARIMA.html.

Currently ARIMARegressor isn’t supported via conda install. It’s recommended that it be installed via PyPI.

Parameters
  • time_index (str) – Specifies the name of the column in X that provides the datetime objects. Defaults to None.

  • trend (str) – Controls the deterministic trend. Options are [‘n’, ‘c’, ‘t’, ‘ct’] where ‘c’ is a constant term, ‘t’ indicates a linear trend, and ‘ct’ is both. Can also be an iterable when defining a polynomial, such as [1, 1, 0, 1].

  • start_p (int) – Minimum Autoregressive order. Defaults to 2.

  • d (int) – Minimum Differencing degree. Defaults to 0.

  • start_q (int) – Minimum Moving Average order. Defaults to 2.

  • max_p (int) – Maximum Autoregressive order. Defaults to 5.

  • max_d (int) – Maximum Differencing degree. Defaults to 2.

  • max_q (int) – Maximum Moving Average order. Defaults to 5.

  • seasonal (boolean) – Whether to fit a seasonal model to ARIMA. Defaults to True.

  • n_jobs (int or None) – Non-negative integer describing level of parallelism used for pipelines. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “start_p”: Integer(1, 3), “d”: Integer(0, 2), “start_q”: Integer(1, 3), “max_p”: Integer(3, 10), “max_d”: Integer(2, 5), “max_q”: Integer(3, 10), “seasonal”: [True, False],}

model_family

ModelFamily.ARIMA

modifies_features

True

modifies_target

False

name

ARIMA Regressor

supported_problem_types

[ProblemTypes.TIME_SERIES_REGRESSION]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns array of 0’s with a length of 1 as feature_importance is not defined for ARIMA regressor.

fit

Fits ARIMA regressor to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using fitted ARIMA regressor.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns array of 0’s with a length of 1 as feature_importance is not defined for ARIMA regressor.

fit(self, X, y=None)[source]

Fits ARIMA regressor to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

Raises

ValueError – If X was passed to fit but not passed in predict.

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X, y=None)[source]

Make predictions using fitted ARIMA regressor.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features].

  • y (pd.Series) – Target data.

Returns

Predicted values.

Return type

pd.Series

Raises

ValueError – If X was passed to fit but not passed in predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.BinaryClassificationPipeline(component_graph, parameters=None, custom_name=None, random_seed=0)[source]

Pipeline subclass for all binary classification pipelines.

Parameters
  • component_graph (ComponentGraph, list, dict) – ComponentGraph instance, list of components in order, or dictionary of components. Accepts strings or ComponentBase subclasses in the list. Note that when duplicate components are specified in a list, the duplicate component names will be modified with the component’s index in the list. For example, the component graph [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names [“Imputer”, “One Hot Encoder”, “Imputer_2”, “Logistic Regression Classifier”]

  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • custom_name (str) – Custom name for the pipeline. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Example

>>> pipeline = BinaryClassificationPipeline(component_graph=["Simple Imputer", "Logistic Regression Classifier"],
...                                         parameters={"Logistic Regression Classifier": {"penalty": "elasticnet",
...                                                                                        "solver": "liblinear"}},
...                                         custom_name="My Binary Pipeline")
...
>>> assert pipeline.custom_name == "My Binary Pipeline"
>>> assert pipeline.component_graph.component_dict.keys() == {'Simple Imputer', 'Logistic Regression Classifier'}

The pipeline parameters will be chosen from the default parameters for every component, unless specific parameters were passed in as they were above.

>>> assert pipeline.parameters == {
...     'Simple Imputer': {'impute_strategy': 'most_frequent', 'fill_value': None},
...     'Logistic Regression Classifier': {'penalty': 'elasticnet',
...                                        'C': 1.0,
...                                        'n_jobs': -1,
...                                        'multi_class': 'auto',
...                                        'solver': 'liblinear'}}

Attributes

problem_type

ProblemTypes.BINARY

Methods

can_tune_threshold_with_objective

Determine whether the threshold of a binary classification pipeline can be tuned.

classes_

Gets the class names for the pipeline. Will return None before pipeline is fit.

clone

Constructs a new pipeline with the same components, parameters, and random seed.

create_objectives

Create objective instances from a list of strings or objective classes.

custom_name

Custom name of the pipeline.

describe

Outputs pipeline details including component parameters.

feature_importance

Importance associated with each feature. Features dropped by the feature selection are excluded.

fit

Build a classification model. For string and categorical targets, classes are sorted by sorted(set(y)) and then are mapped to values between 0 and n_classes-1.

get_component

Returns component by name.

get_hyperparameter_ranges

Returns hyperparameter ranges from all components as a dictionary.

graph

Generate an image representing the pipeline graph.

graph_feature_importance

Generate a bar graph of the pipeline’s feature importance.

graph_json

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

inverse_transform

Apply component inverse_transform methods to estimator predictions in reverse order.

load

Loads pipeline at file path.

model_family

Returns model family of this pipeline.

name

Name of the pipeline.

new

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

optimize_threshold

Optimize the pipeline threshold given the objective to use. Only used for binary problems with objectives whose thresholds can be tuned.

parameters

Parameter dictionary for this pipeline.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels. Assumes that the column at index 1 represents the positive label case.

save

Saves pipeline at file path.

score

Evaluate model performance on objectives.

summary

A short summary of the pipeline structure, describing the list of components used.

threshold

Threshold used to make a prediction. Defaults to None.

transform

Transform the input.

transform_all_but_final

Transforms the data by applying all pre-processing components.

can_tune_threshold_with_objective(self, objective)

Determine whether the threshold of a binary classification pipeline can be tuned.

Parameters

objective (ObjectiveBase) – Primary AutoMLSearch objective.

Returns

True if the pipeline threshold can be tuned.

Return type

bool

property classes_(self)

Gets the class names for the pipeline. Will return None before pipeline is fit.

clone(self)

Constructs a new pipeline with the same components, parameters, and random seed.

Returns

A new instance of this pipeline with identical components, parameters, and random seed.

static create_objectives(objectives)

Create objective instances from a list of strings or objective classes.

property custom_name(self)

Custom name of the pipeline.

describe(self, return_dict=False)

Outputs pipeline details including component parameters.

Parameters

return_dict (bool) – If True, return dictionary of information about pipeline. Defaults to False.

Returns

Dictionary of all component parameters if return_dict is True, else None.

Return type

dict

property feature_importance(self)

Importance associated with each feature. Features dropped by the feature selection are excluded.

Returns

Feature names and their corresponding importance

Return type

pd.DataFrame

fit(self, X, y)

Build a classification model. For string and categorical targets, classes are sorted by sorted(set(y)) and then are mapped to values between 0 and n_classes-1.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, np.ndarray) – The target training labels of length [n_samples]

Returns

self

get_component(self, name)

Returns component by name.

Parameters

name (str) – Name of component.

Returns

Component to return

Return type

Component

get_hyperparameter_ranges(self, custom_hyperparameters)

Returns hyperparameter ranges from all components as a dictionary.

Parameters

custom_hyperparameters (dict) – Custom hyperparameters for the pipeline.

Returns

Dictionary of hyperparameter ranges for each component in the pipeline.

Return type

dict

graph(self, filepath=None)

Generate an image representing the pipeline graph.

Parameters

filepath (str, optional) – Path to where the graph should be saved. If set to None (as by default), the graph will not be saved.

Returns

Graph object that can be directly displayed in Jupyter notebooks.

Return type

graphviz.Digraph

Raises
  • RuntimeError – If graphviz is not installed.

  • ValueError – If path is not writeable.

graph_feature_importance(self, importance_threshold=0)

Generate a bar graph of the pipeline’s feature importance.

Parameters

importance_threshold (float, optional) – If provided, graph features with a permutation importance whose absolute value is larger than importance_threshold. Defaults to zero.

Returns

A bar graph showing features and their corresponding importance.

Return type

plotly.Figure

Raises

ValueError – If importance threshold is not valid.

graph_json(self)

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

x_edges specifies from which component feature data is being passed. y_edges specifies from which component target data is being passed. This can be used to build graphs across a variety of visualization tools. Template: {“Nodes”: {“component_name”: {“Name”: class_name, “Parameters”: parameters_attributes}, …}}, “x_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …], “y_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …]}

Returns

A serialized JSON representation of a DAG structure.

Return type

dag_json (str)

inverse_transform(self, y)

Apply component inverse_transform methods to estimator predictions in reverse order.

Components that implement inverse_transform are PolynomialDetrender, LogTransformer, LabelEncoder (tbd).

Parameters

y (pd.Series) – Final component features.

Returns

The inverse transform of the target.

Return type

pd.Series

static load(file_path)

Loads pipeline at file path.

Parameters

file_path (str) – Location to load file.

Returns

PipelineBase object

property model_family(self)

Returns model family of this pipeline.

property name(self)

Name of the pipeline.

new(self, parameters, random_seed=0)

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

Parameters
  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Returns

A new instance of this pipeline with identical components.

optimize_threshold(self, X, y, y_pred_proba, objective)

Optimize the pipeline threshold given the objective to use. Only used for binary problems with objectives whose thresholds can be tuned.

Parameters
  • X (pd.DataFrame) – Input features.

  • y (pd.Series) – Input target values.

  • y_pred_proba (pd.Series) – The predicted probabilities of the target outputted by the pipeline.

  • objective (ObjectiveBase) – The objective to threshold with. Must have a tunable threshold.

Raises

ValueError – If objective is not optimizable.

property parameters(self)

Parameter dictionary for this pipeline.

Returns

Dictionary of all component parameters.

Return type

dict

predict(self, X, objective=None, X_train=None, y_train=None)

Make predictions using selected features.

Note: we cast y as ints first to address boolean values that may be returned from calculating predictions which we would not be able to otherwise transform if we originally had integer targets.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features].

  • objective (Object or string) – The objective to use to make predictions.

  • X_train (pd.DataFrame) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series) – Training labels. Ignored. Only used for time series.

Returns

Estimated labels.

Return type

pd.Series

predict_proba(self, X, X_train=None, y_train=None)[source]

Make probability estimates for labels. Assumes that the column at index 1 represents the positive label case.

Parameters
  • X (pd.DataFrame or np.ndarray) – Data of shape [n_samples, n_features]

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Probability estimates

Return type

pd.Series

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves pipeline at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

score(self, X, y, objectives, X_train=None, y_train=None)

Evaluate model performance on objectives.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features]

  • y (pd.Series) – True labels of length [n_samples]

  • objectives (list) – List of objectives to score

  • X_train (pd.DataFrame) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series) – Training labels. Ignored. Only used for time series.

Returns

Ordered dictionary of objective scores.

Return type

dict

property summary(self)

A short summary of the pipeline structure, describing the list of components used.

Example: Logistic Regression Classifier w/ Simple Imputer + One Hot Encoder

Returns

A string describing the pipeline structure.

property threshold(self)

Threshold used to make a prediction. Defaults to None.

transform(self, X, y=None)

Transform the input.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • y (pd.Series) – The target data of length [n_samples]. Defaults to None.

Returns

Transformed output.

Return type

pd.DataFrame

transform_all_but_final(self, X, y=None, X_train=None, y_train=None)

Transforms the data by applying all pre-processing components.

Parameters
  • X (pd.DataFrame) – Input data to the pipeline to transform.

  • y (pd.Series or None) – Targets corresponding to X. Optional.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Only used for time series.

Returns

New transformed features.

Return type

pd.DataFrame

class evalml.pipelines.CatBoostClassifier(n_estimators=10, eta=0.03, max_depth=6, bootstrap_type=None, silent=True, allow_writing_files=False, random_seed=0, n_jobs=- 1, **kwargs)[source]

CatBoost Classifier, a classifier that uses gradient-boosting on decision trees. CatBoost is an open-source library and natively supports categorical features.

For more information, check out https://catboost.ai/

Parameters
  • n_estimators (float) – The maximum number of trees to build. Defaults to 10.

  • eta (float) – The learning rate. Defaults to 0.03.

  • max_depth (int) – The maximum tree depth for base learners. Defaults to 6.

  • bootstrap_type (string) – Defines the method for sampling the weights of objects. Available methods are ‘Bayesian’, ‘Bernoulli’, ‘MVS’. Defaults to None.

  • silent (boolean) – Whether to use the “silent” logging mode. Defaults to True.

  • allow_writing_files (boolean) – Whether to allow writing snapshot files while training. Defaults to False.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_estimators”: Integer(4, 100), “eta”: Real(0.000001, 1), “max_depth”: Integer(4, 10),}

model_family

ModelFamily.CATBOOST

modifies_features

True

modifies_target

False

name

CatBoost Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance of fitted CatBoost classifier.

fit

Fits CatBoost classifier component to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using the fitted CatBoost classifier.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance of fitted CatBoost classifier.

fit(self, X, y=None)[source]

Fits CatBoost classifier component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)[source]

Make predictions using the fitted CatBoost classifier.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.DataFrame

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.CatBoostRegressor(n_estimators=10, eta=0.03, max_depth=6, bootstrap_type=None, silent=False, allow_writing_files=False, random_seed=0, n_jobs=- 1, **kwargs)[source]

CatBoost Regressor, a regressor that uses gradient-boosting on decision trees. CatBoost is an open-source library and natively supports categorical features.

For more information, check out https://catboost.ai/

Parameters
  • n_estimators (float) – The maximum number of trees to build. Defaults to 10.

  • eta (float) – The learning rate. Defaults to 0.03.

  • max_depth (int) – The maximum tree depth for base learners. Defaults to 6.

  • bootstrap_type (string) – Defines the method for sampling the weights of objects. Available methods are ‘Bayesian’, ‘Bernoulli’, ‘MVS’. Defaults to None.

  • silent (boolean) – Whether to use the “silent” logging mode. Defaults to True.

  • allow_writing_files (boolean) – Whether to allow writing snapshot files while training. Defaults to False.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_estimators”: Integer(4, 100), “eta”: Real(0.000001, 1), “max_depth”: Integer(4, 10),}

model_family

ModelFamily.CATBOOST

modifies_features

True

modifies_target

False

name

CatBoost Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance of fitted CatBoost regressor.

fit

Fits CatBoost regressor component to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance of fitted CatBoost regressor.

fit(self, X, y=None)[source]

Fits CatBoost regressor component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.ClassificationPipeline(component_graph, parameters=None, custom_name=None, random_seed=0)[source]

Pipeline subclass for all classification pipelines.

Parameters
  • component_graph (list or dict) – List of components in order. Accepts strings or ComponentBase subclasses in the list. Note that when duplicate components are specified in a list, the duplicate component names will be modified with the component’s index in the list. For example, the component graph [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names [“Imputer”, “One Hot Encoder”, “Imputer_2”, “Logistic Regression Classifier”]

  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • custom_name (str) – Custom name for the pipeline. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

problem_type

None

Methods

can_tune_threshold_with_objective

Determine whether the threshold of a binary classification pipeline can be tuned.

classes_

Gets the class names for the pipeline. Will return None before pipeline is fit.

clone

Constructs a new pipeline with the same components, parameters, and random seed.

create_objectives

Create objective instances from a list of strings or objective classes.

custom_name

Custom name of the pipeline.

describe

Outputs pipeline details including component parameters.

feature_importance

Importance associated with each feature. Features dropped by the feature selection are excluded.

fit

Build a classification model. For string and categorical targets, classes are sorted by sorted(set(y)) and then are mapped to values between 0 and n_classes-1.

get_component

Returns component by name.

get_hyperparameter_ranges

Returns hyperparameter ranges from all components as a dictionary.

graph

Generate an image representing the pipeline graph.

graph_feature_importance

Generate a bar graph of the pipeline’s feature importance.

graph_json

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

inverse_transform

Apply component inverse_transform methods to estimator predictions in reverse order.

load

Loads pipeline at file path.

model_family

Returns model family of this pipeline.

name

Name of the pipeline.

new

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

parameters

Parameter dictionary for this pipeline.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves pipeline at file path.

score

Evaluate model performance on objectives.

summary

A short summary of the pipeline structure, describing the list of components used.

transform

Transform the input.

transform_all_but_final

Transforms the data by applying all pre-processing components.

can_tune_threshold_with_objective(self, objective)

Determine whether the threshold of a binary classification pipeline can be tuned.

Parameters

objective (ObjectiveBase) – Primary AutoMLSearch objective.

Returns

True if the pipeline threshold can be tuned.

Return type

bool

property classes_(self)

Gets the class names for the pipeline. Will return None before pipeline is fit.

clone(self)

Constructs a new pipeline with the same components, parameters, and random seed.

Returns

A new instance of this pipeline with identical components, parameters, and random seed.

static create_objectives(objectives)

Create objective instances from a list of strings or objective classes.

property custom_name(self)

Custom name of the pipeline.

describe(self, return_dict=False)

Outputs pipeline details including component parameters.

Parameters

return_dict (bool) – If True, return dictionary of information about pipeline. Defaults to False.

Returns

Dictionary of all component parameters if return_dict is True, else None.

Return type

dict

property feature_importance(self)

Importance associated with each feature. Features dropped by the feature selection are excluded.

Returns

Feature names and their corresponding importance

Return type

pd.DataFrame

fit(self, X, y)[source]

Build a classification model. For string and categorical targets, classes are sorted by sorted(set(y)) and then are mapped to values between 0 and n_classes-1.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, np.ndarray) – The target training labels of length [n_samples]

Returns

self

get_component(self, name)

Returns component by name.

Parameters

name (str) – Name of component.

Returns

Component to return

Return type

Component

get_hyperparameter_ranges(self, custom_hyperparameters)

Returns hyperparameter ranges from all components as a dictionary.

Parameters

custom_hyperparameters (dict) – Custom hyperparameters for the pipeline.

Returns

Dictionary of hyperparameter ranges for each component in the pipeline.

Return type

dict

graph(self, filepath=None)

Generate an image representing the pipeline graph.

Parameters

filepath (str, optional) – Path to where the graph should be saved. If set to None (as by default), the graph will not be saved.

Returns

Graph object that can be directly displayed in Jupyter notebooks.

Return type

graphviz.Digraph

Raises
  • RuntimeError – If graphviz is not installed.

  • ValueError – If path is not writeable.

graph_feature_importance(self, importance_threshold=0)

Generate a bar graph of the pipeline’s feature importance.

Parameters

importance_threshold (float, optional) – If provided, graph features with a permutation importance whose absolute value is larger than importance_threshold. Defaults to zero.

Returns

A bar graph showing features and their corresponding importance.

Return type

plotly.Figure

Raises

ValueError – If importance threshold is not valid.

graph_json(self)

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

x_edges specifies from which component feature data is being passed. y_edges specifies from which component target data is being passed. This can be used to build graphs across a variety of visualization tools. Template: {“Nodes”: {“component_name”: {“Name”: class_name, “Parameters”: parameters_attributes}, …}}, “x_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …], “y_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …]}

Returns

A serialized JSON representation of a DAG structure.

Return type

dag_json (str)

inverse_transform(self, y)

Apply component inverse_transform methods to estimator predictions in reverse order.

Components that implement inverse_transform are PolynomialDetrender, LogTransformer, LabelEncoder (tbd).

Parameters

y (pd.Series) – Final component features.

Returns

The inverse transform of the target.

Return type

pd.Series

static load(file_path)

Loads pipeline at file path.

Parameters

file_path (str) – Location to load file.

Returns

PipelineBase object

property model_family(self)

Returns model family of this pipeline.

property name(self)

Name of the pipeline.

new(self, parameters, random_seed=0)

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

Parameters
  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Returns

A new instance of this pipeline with identical components.

property parameters(self)

Parameter dictionary for this pipeline.

Returns

Dictionary of all component parameters.

Return type

dict

predict(self, X, objective=None, X_train=None, y_train=None)[source]

Make predictions using selected features.

Note: we cast y as ints first to address boolean values that may be returned from calculating predictions which we would not be able to otherwise transform if we originally had integer targets.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features].

  • objective (Object or string) – The objective to use to make predictions.

  • X_train (pd.DataFrame) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series) – Training labels. Ignored. Only used for time series.

Returns

Estimated labels.

Return type

pd.Series

predict_proba(self, X, X_train=None, y_train=None)[source]

Make probability estimates for labels.

Parameters
  • X (pd.DataFrame or np.ndarray) – Data of shape [n_samples, n_features]

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Probability estimates

Return type

pd.DataFrame

Raises

ValueError – If final component is not an estimator.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves pipeline at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

score(self, X, y, objectives, X_train=None, y_train=None)[source]

Evaluate model performance on objectives.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features]

  • y (pd.Series) – True labels of length [n_samples]

  • objectives (list) – List of objectives to score

  • X_train (pd.DataFrame) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series) – Training labels. Ignored. Only used for time series.

Returns

Ordered dictionary of objective scores.

Return type

dict

property summary(self)

A short summary of the pipeline structure, describing the list of components used.

Example: Logistic Regression Classifier w/ Simple Imputer + One Hot Encoder

Returns

A string describing the pipeline structure.

transform(self, X, y=None)

Transform the input.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • y (pd.Series) – The target data of length [n_samples]. Defaults to None.

Returns

Transformed output.

Return type

pd.DataFrame

transform_all_but_final(self, X, y=None, X_train=None, y_train=None)

Transforms the data by applying all pre-processing components.

Parameters
  • X (pd.DataFrame) – Input data to the pipeline to transform.

  • y (pd.Series or None) – Targets corresponding to X. Optional.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Only used for time series.

Returns

New transformed features.

Return type

pd.DataFrame

class evalml.pipelines.ComponentGraph(component_dict=None, random_seed=0)[source]

Component graph for a pipeline as a directed acyclic graph (DAG).

Parameters
  • component_dict (dict) – A dictionary which specifies the components and edges between components that should be used to create the component graph. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Examples

>>> component_dict = {'Imputer': ['Imputer', 'X', 'y'],
...                   'Logistic Regression': ['Logistic Regression Classifier', 'Imputer.x', 'y']}
>>> component_graph = ComponentGraph(component_dict)
>>> assert component_graph.compute_order == ['Imputer', 'Logistic Regression']
...
...
>>> component_dict = {'Imputer': ['Imputer', 'X', 'y'],
...                   'OHE': ['One Hot Encoder', 'Imputer.x', 'y'],
...                   'estimator_1': ['Random Forest Classifier', 'OHE.x', 'y'],
...                   'estimator_2': ['Decision Tree Classifier', 'OHE.x', 'y'],
...                   'final': ['Logistic Regression Classifier', 'estimator_1.x', 'estimator_2.x', 'y']}
>>> component_graph = ComponentGraph(component_dict)

The default parameters for every component in the component graph.

>>> assert component_graph.default_parameters == {
...     'Imputer': {'categorical_impute_strategy': 'most_frequent',
...                 'numeric_impute_strategy': 'mean',
...                 'categorical_fill_value': None,
...                 'numeric_fill_value': None},
...     'One Hot Encoder': {'top_n': 10,
...                         'features_to_encode': None,
...                         'categories': None,
...                         'drop': 'if_binary',
...                         'handle_unknown': 'ignore',
...                         'handle_missing': 'error'},
...     'Random Forest Classifier': {'n_estimators': 100,
...                                  'max_depth': 6,
...                                  'n_jobs': -1},
...     'Decision Tree Classifier': {'criterion': 'gini',
...                                  'max_features': 'auto',
...                                  'max_depth': 6,
...                                  'min_samples_split': 2,
...                                  'min_weight_fraction_leaf': 0.0},
...     'Logistic Regression Classifier': {'penalty': 'l2',
...                                        'C': 1.0,
...                                        'n_jobs': -1,
...                                        'multi_class': 'auto',
...                                        'solver': 'lbfgs'}}

Methods

compute_order

The order that components will be computed or called in.

default_parameters

The default parameter dictionary for this pipeline.

describe

Outputs component graph details including component parameters.

fit

Fit each component in the graph.

fit_and_transform_all_but_final

Fit and transform all components save the final one, usually an estimator.

generate_order

Regenerated the topologically sorted order of the graph.

get_component

Retrieves a single component object from the graph.

get_estimators

Gets a list of all the estimator components within this graph.

get_inputs

Retrieves all inputs for a given component.

get_last_component

Retrieves the component that is computed last in the graph, usually the final estimator.

graph

Generate an image representing the component graph.

instantiate

Instantiates all uninstantiated components within the graph using the given parameters. An error will be raised if a component is already instantiated but the parameters dict contains arguments for that component.

inverse_transform

Apply component inverse_transform methods to estimator predictions in reverse order.

predict

Make predictions using selected features.

transform

Transform the input using the component graph.

transform_all_but_final

Transform all components save the final one, and gathers the data from any number of parents to get all the information that should be fed to the final component.

property compute_order(self)

The order that components will be computed or called in.

property default_parameters(self)

The default parameter dictionary for this pipeline.

Returns

Dictionary of all component default parameters.

Return type

dict

describe(self, return_dict=False)[source]

Outputs component graph details including component parameters.

Parameters

return_dict (bool) – If True, return dictionary of information about component graph. Defaults to False.

Returns

Dictionary of all component parameters if return_dict is True, else None

Return type

dict

fit(self, X, y)[source]

Fit each component in the graph.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

fit_and_transform_all_but_final(self, X, y)[source]

Fit and transform all components save the final one, usually an estimator.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

Transformed features and target.

Return type

Tuple (pd.DataFrame, pd.Series)

classmethod generate_order(cls, component_dict)[source]

Regenerated the topologically sorted order of the graph.

get_component(self, component_name)[source]

Retrieves a single component object from the graph.

Parameters

component_name (str) – Name of the component to retrieve

Returns

ComponentBase object

Raises

ValueError – If the component is not in the graph.

get_estimators(self)[source]

Gets a list of all the estimator components within this graph.

Returns

All estimator objects within the graph.

Return type

list

Raises

ValueError – If the component graph is not yet instantiated.

get_inputs(self, component_name)[source]

Retrieves all inputs for a given component.

Parameters

component_name (str) – Name of the component to look up.

Returns

List of inputs for the component to use.

Return type

list[str]

Raises

ValueError – If the component is not in the graph.

get_last_component(self)[source]

Retrieves the component that is computed last in the graph, usually the final estimator.

Returns

ComponentBase object

Raises

ValueError – If the component graph has no edges.

graph(self, name=None, graph_format=None)[source]

Generate an image representing the component graph.

Parameters
  • name (str) – Name of the graph. Defaults to None.

  • graph_format (str) – file format to save the graph in. Defaults to None.

Returns

Graph object that can be directly displayed in Jupyter notebooks.

Return type

graphviz.Digraph

Raises

RuntimeError – If graphviz is not installed.

instantiate(self, parameters=None)[source]

Instantiates all uninstantiated components within the graph using the given parameters. An error will be raised if a component is already instantiated but the parameters dict contains arguments for that component.

Parameters

parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary {} or None implies using all default values for component parameters. If a component in the component graph is already instantiated, it will not use any of its parameters defined in this dictionary. Defaults to None.

Returns

self

Raises

ValueError – If component graph is already instantiated or if a component errored while instantiating.

inverse_transform(self, y)[source]

Apply component inverse_transform methods to estimator predictions in reverse order.

Components that implement inverse_transform are PolynomialDetrender, LabelEncoder (tbd).

Parameters

y – (pd.Series): Final component features.

Returns

The target with inverse transformation applied.

Return type

pd.Series

predict(self, X)[source]

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Input features of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

ValueError – If final component is not an Estimator.

transform(self, X, y=None)[source]

Transform the input using the component graph.

Parameters
  • X (pd.DataFrame) – Input features of shape [n_samples, n_features].

  • y (pd.Series) – The target data of length [n_samples]. Defaults to None.

Returns

Transformed output.

Return type

pd.DataFrame

Raises

ValueError – If final component is not a Transformer.

transform_all_but_final(self, X, y=None)[source]

Transform all components save the final one, and gathers the data from any number of parents to get all the information that should be fed to the final component.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples]. Defaults to None.

Returns

Transformed values.

Return type

pd.DataFrame

class evalml.pipelines.DecisionTreeClassifier(criterion='gini', max_features='auto', max_depth=6, min_samples_split=2, min_weight_fraction_leaf=0.0, random_seed=0, **kwargs)[source]

Decision Tree Classifier.

Parameters
  • criterion ({"gini", "entropy"}) – The function to measure the quality of a split. Supported criteria are “gini” for the Gini impurity and “entropy” for the information gain. Defaults to “gini”.

  • max_features (int, float or {"auto", "sqrt", "log2"}) –

    The number of features to consider when looking for the best split:

    • If int, then consider max_features features at each split.

    • If float, then max_features is a fraction and int(max_features * n_features) features are considered at each split.

    • If “auto”, then max_features=sqrt(n_features).

    • If “sqrt”, then max_features=sqrt(n_features).

    • If “log2”, then max_features=log2(n_features).

    • If None, then max_features = n_features.

    The search for a split does not stop until at least one valid partition of the node samples is found, even if it requires to effectively inspect more than max_features features. Defaults to “auto”.

  • max_depth (int) – The maximum depth of the tree. Defaults to 6.

  • min_samples_split (int or float) –

    The minimum number of samples required to split an internal node:

    • If int, then consider min_samples_split as the minimum number.

    • If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.

    Defaults to 2.

  • min_weight_fraction_leaf (float) – The minimum weighted fraction of the sum total of weights (of all the input samples) required to be at a leaf node. Defaults to 0.0.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “criterion”: [“gini”, “entropy”], “max_features”: [“auto”, “sqrt”, “log2”], “max_depth”: Integer(4, 10),}

model_family

ModelFamily.DECISION_TREE

modifies_features

True

modifies_target

False

name

Decision Tree Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.DecisionTreeRegressor(criterion='mse', max_features='auto', max_depth=6, min_samples_split=2, min_weight_fraction_leaf=0.0, random_seed=0, **kwargs)[source]

Decision Tree Regressor.

Parameters
  • criterion ({"mse", "friedman_mse", "mae", "poisson"}) –

    The function to measure the quality of a split. Supported criteria are:

    • ”mse” for the mean squared error, which is equal to variance reduction as feature selection criterion and minimizes the L2 loss using the mean of each terminal node

    • ”friedman_mse”, which uses mean squared error with Friedman”s improvement score for potential splits

    • ”mae” for the mean absolute error, which minimizes the L1 loss using the median of each terminal node,

    • ”poisson” which uses reduction in Poisson deviance to find splits.

  • max_features (int, float or {"auto", "sqrt", "log2"}) –

    The number of features to consider when looking for the best split:

    • If int, then consider max_features features at each split.

    • If float, then max_features is a fraction and int(max_features * n_features) features are considered at each split.

    • If “auto”, then max_features=sqrt(n_features).

    • If “sqrt”, then max_features=sqrt(n_features).

    • If “log2”, then max_features=log2(n_features).

    • If None, then max_features = n_features.

    The search for a split does not stop until at least one valid partition of the node samples is found, even if it requires to effectively inspect more than max_features features.

  • max_depth (int) – The maximum depth of the tree. Defaults to 6.

  • min_samples_split (int or float) –

    The minimum number of samples required to split an internal node:

    • If int, then consider min_samples_split as the minimum number.

    • If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.

    Defaults to 2.

  • min_weight_fraction_leaf (float) – The minimum weighted fraction of the sum total of weights (of all the input samples) required to be at a leaf node. Defaults to 0.0.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “criterion”: [“mse”, “friedman_mse”, “mae”], “max_features”: [“auto”, “sqrt”, “log2”], “max_depth”: Integer(4, 10),}

model_family

ModelFamily.DECISION_TREE

modifies_features

True

modifies_target

False

name

Decision Tree Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.DFSTransformer(index='index', features=None, random_seed=0, **kwargs)[source]

Featuretools DFS component that generates features for the input features.

Parameters
  • index (string) – The name of the column that contains the indices. If no column with this name exists, then featuretools.EntitySet() creates a column with this name to serve as the index column. Defaults to ‘index’.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

  • features (list) – List of features to run DFS on. Defaults to None. Features will only be computed if the columns used by the feature exist in the input and if the feature itself is not in input.

Attributes

hyperparameter_ranges

{}

modifies_features

True

modifies_target

False

name

DFS Transformer

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits the DFSTransformer Transformer component.

fit_transform

Fits on X and transforms X.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Computes the feature matrix for the input X using featuretools’ dfs algorithm.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)[source]

Fits the DFSTransformer Transformer component.

Parameters
  • X (pd.DataFrame, np.array) – The input data to transform, of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

fit_transform(self, X, y=None)

Fits on X and transforms X.

Parameters
  • X (pd.DataFrame) – Data to fit and transform.

  • y (pd.Series) – Target data.

Returns

Transformed X.

Return type

pd.DataFrame

Raises

MethodPropertyNotFoundError – If transformer does not have a transform method or a component_obj that implements transform.

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)[source]

Computes the feature matrix for the input X using featuretools’ dfs algorithm.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data to transform. Has shape [n_samples, n_features]

  • y (pd.Series, optional) – Ignored.

Returns

Feature matrix

Return type

pd.DataFrame

class evalml.pipelines.ElasticNetClassifier(penalty='elasticnet', C=1.0, l1_ratio=0.15, multi_class='auto', solver='saga', n_jobs=- 1, random_seed=0, **kwargs)[source]

Elastic Net Classifier. Uses Logistic Regression with elasticnet penalty as the base estimator.

Parameters
  • penalty ({"l1", "l2", "elasticnet", "none"}) – The norm used in penalization. Defaults to “elasticnet”.

  • C (float) – Inverse of regularization strength. Must be a positive float. Defaults to 1.0.

  • l1_ratio (float) – The mixing parameter, with 0 <= l1_ratio <= 1. Only used if penalty=’elasticnet’. Setting l1_ratio=0 is equivalent to using penalty=’l2’, while setting l1_ratio=1 is equivalent to using penalty=’l1’. For 0 < l1_ratio <1, the penalty is a combination of L1 and L2. Defaults to 0.15.

  • multi_class ({"auto", "ovr", "multinomial"}) – If the option chosen is “ovr”, then a binary problem is fit for each label. For “multinomial” the loss minimised is the multinomial loss fit across the entire probability distribution, even when the data is binary. “multinomial” is unavailable when solver=”liblinear”. “auto” selects “ovr” if the data is binary, or if solver=”liblinear”, and otherwise selects “multinomial”. Defaults to “auto”.

  • solver ({"newton-cg", "lbfgs", "liblinear", "sag", "saga"}) –

    Algorithm to use in the optimization problem. For small datasets, “liblinear” is a good choice, whereas “sag” and “saga” are faster for large ones. For multiclass problems, only “newton-cg”, “sag”, “saga” and “lbfgs” handle multinomial loss; “liblinear” is limited to one-versus-rest schemes.

    • ”newton-cg”, “lbfgs”, “sag” and “saga” handle L2 or no penalty

    • ”liblinear” and “saga” also handle L1 penalty

    • ”saga” also supports “elasticnet” penalty

    • ”liblinear” does not support setting penalty=’none’

    Defaults to “saga”.

  • n_jobs (int) – Number of parallel threads used to run xgboost. Note that creating thread contention will significantly slow down the algorithm. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “C”: Real(0.01, 10), “l1_ratio”: Real(0, 1)}

model_family

ModelFamily.LINEAR_MODEL

modifies_features

True

modifies_target

False

name

Elastic Net Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance for fitted ElasticNet classifier.

fit

Fits ElasticNet classifier component to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance for fitted ElasticNet classifier.

fit(self, X, y)[source]

Fits ElasticNet classifier component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.ElasticNetRegressor(alpha=0.0001, l1_ratio=0.15, max_iter=1000, normalize=False, random_seed=0, **kwargs)[source]

Elastic Net Regressor.

Parameters
  • alpha (float) – Constant that multiplies the penalty terms. Defaults to 0.0001.

  • l1_ratio (float) – The mixing parameter, with 0 <= l1_ratio <= 1. Only used if penalty=’elasticnet’. Setting l1_ratio=0 is equivalent to using penalty=’l2’, while setting l1_ratio=1 is equivalent to using penalty=’l1’. For 0 < l1_ratio <1, the penalty is a combination of L1 and L2. Defaults to 0.15.

  • max_iter (int) – The maximum number of iterations. Defaults to 1000.

  • normalize (boolean) – If True, the regressors will be normalized before regression by subtracting the mean and dividing by the l2-norm. Defaults to False.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “alpha”: Real(0, 1), “l1_ratio”: Real(0, 1),}

model_family

ModelFamily.LINEAR_MODEL

modifies_features

True

modifies_target

False

name

Elastic Net Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance for fitted ElasticNet regressor.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance for fitted ElasticNet regressor.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.Estimator(parameters=None, component_obj=None, random_seed=0, **kwargs)[source]

A component that fits and predicts given data.

To implement a new Estimator, define your own class which is a subclass of Estimator, including a name and a list of acceptable ranges for any parameters to be tuned during the automl search (hyperparameters). Define an __init__ method which sets up any necessary state and objects. Make sure your __init__ only uses standard keyword arguments and calls super().__init__() with a parameters dict. You may also override the fit, transform, fit_transform and other methods in this class if appropriate.

To see some examples, check out the definitions of any Estimator component subclass.

Parameters
  • parameters (dict) – Dictionary of parameters for the component. Defaults to None.

  • component_obj (obj) – Third-party objects useful in component implementation. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

model_family

ModelFamily.NONE

modifies_features

True

modifies_target

False

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

model_family

ModelFamily.NONE

name

Returns string name of this component.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

supported_problem_types

Problem types this estimator supports.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)[source]

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

property model_family(cls)

Returns ModelFamily of this component.

property name(cls)

Returns string name of this component.

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)[source]

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)[source]

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

property supported_problem_types(cls)

Problem types this estimator supports.

class evalml.pipelines.ExtraTreesClassifier(n_estimators=100, max_features='auto', max_depth=6, min_samples_split=2, min_weight_fraction_leaf=0.0, n_jobs=- 1, random_seed=0, **kwargs)[source]

Extra Trees Classifier.

Parameters
  • n_estimators (float) – The number of trees in the forest. Defaults to 100.

  • max_features (int, float or {"auto", "sqrt", "log2"}) –

    The number of features to consider when looking for the best split:

    • If int, then consider max_features features at each split.

    • If float, then max_features is a fraction and int(max_features * n_features) features are considered at each split.

    • If “auto”, then max_features=sqrt(n_features).

    • If “sqrt”, then max_features=sqrt(n_features).

    • If “log2”, then max_features=log2(n_features).

    • If None, then max_features = n_features.

    The search for a split does not stop until at least one valid partition of the node samples is found, even if it requires to effectively inspect more than max_features features. Defaults to “auto”.

  • max_depth (int) – The maximum depth of the tree. Defaults to 6.

  • min_samples_split (int or float) –

    The minimum number of samples required to split an internal node:

    • If int, then consider min_samples_split as the minimum number.

    • If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.

  • to 2. (Defaults) –

  • min_weight_fraction_leaf (float) – The minimum weighted fraction of the sum total of weights (of all the input samples) required to be at a leaf node. Defaults to 0.0.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_estimators”: Integer(10, 1000), “max_features”: [“auto”, “sqrt”, “log2”], “max_depth”: Integer(4, 10),}

model_family

ModelFamily.EXTRA_TREES

modifies_features

True

modifies_target

False

name

Extra Trees Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.ExtraTreesRegressor(n_estimators=100, max_features='auto', max_depth=6, min_samples_split=2, min_weight_fraction_leaf=0.0, n_jobs=- 1, random_seed=0, **kwargs)[source]

Extra Trees Regressor.

Parameters
  • n_estimators (float) – The number of trees in the forest. Defaults to 100.

  • max_features (int, float or {"auto", "sqrt", "log2"}) –

    The number of features to consider when looking for the best split:

    • If int, then consider max_features features at each split.

    • If float, then max_features is a fraction and int(max_features * n_features) features are considered at each split.

    • If “auto”, then max_features=sqrt(n_features).

    • If “sqrt”, then max_features=sqrt(n_features).

    • If “log2”, then max_features=log2(n_features).

    • If None, then max_features = n_features.

    The search for a split does not stop until at least one valid partition of the node samples is found, even if it requires to effectively inspect more than max_features features. Defaults to “auto”.

  • max_depth (int) – The maximum depth of the tree. Defaults to 6.

  • min_samples_split (int or float) –

    The minimum number of samples required to split an internal node:

    • If int, then consider min_samples_split as the minimum number.

    • If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.

  • to 2. (Defaults) –

  • min_weight_fraction_leaf (float) – The minimum weighted fraction of the sum total of weights (of all the input samples) required to be at a leaf node. Defaults to 0.0.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_estimators”: Integer(10, 1000), “max_features”: [“auto”, “sqrt”, “log2”], “max_depth”: Integer(4, 10),}

model_family

ModelFamily.EXTRA_TREES

modifies_features

True

modifies_target

False

name

Extra Trees Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.FeatureSelector(parameters=None, component_obj=None, random_seed=0, **kwargs)[source]

Selects top features based on importance weights.

Parameters
  • parameters (dict) – Dictionary of parameters for the component. Defaults to None.

  • component_obj (obj) – Third-party objects useful in component implementation. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

modifies_features

True

modifies_target

False

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits component to data.

fit_transform

Fit and transform data using the feature selector.

get_names

Get names of selected features.

load

Loads component at file path.

name

Returns string name of this component.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Transforms input data by selecting features. If the component_obj does not have a transform method, will raise an MethodPropertyNotFoundError exception.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)

Fits component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, optional) – The target training data of length [n_samples]

Returns

self

Raises

MethodPropertyNotFoundError – If component does not have a fit method or a component_obj that implements fit.

fit_transform(self, X, y=None)[source]

Fit and transform data using the feature selector.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

Transformed data.

Return type

pd.DataFrame

get_names(self)[source]

Get names of selected features.

Returns

List of the names of features selected.

Return type

list[str]

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

property name(cls)

Returns string name of this component.

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)[source]

Transforms input data by selecting features. If the component_obj does not have a transform method, will raise an MethodPropertyNotFoundError exception.

Parameters
  • X (pd.DataFrame) – Data to transform.

  • y (pd.Series, optional) – Target data. Ignored.

Returns

Transformed X

Return type

pd.DataFrame

Raises

MethodPropertyNotFoundError – If feature selector does not have a transform method or a component_obj that implements transform

class evalml.pipelines.KNeighborsClassifier(n_neighbors=5, weights='uniform', algorithm='auto', leaf_size=30, p=2, random_seed=0, **kwargs)[source]

K-Nearest Neighbors Classifier.

Parameters
  • n_neighbors (int) – Number of neighbors to use by default. Defaults to 5.

  • weights ({‘uniform’, ‘distance’} or callable) –

    Weight function used in prediction. Can be:

    • ‘uniform’ : uniform weights. All points in each neighborhood are weighted equally.

    • ‘distance’ : weight points by the inverse of their distance. in this case, closer neighbors of a query point will have a greater influence than neighbors which are further away.

    • [callable] : a user-defined function which accepts an array of distances, and returns an array of the same shape containing the weights.

    Defaults to “uniform”.

  • algorithm ({‘auto’, ‘ball_tree’, ‘kd_tree’, ‘brute’}) –

    Algorithm used to compute the nearest neighbors:

    • ‘ball_tree’ will use BallTree

    • ‘kd_tree’ will use KDTree

    • ‘brute’ will use a brute-force search.

    ‘auto’ will attempt to decide the most appropriate algorithm based on the values passed to fit method. Defaults to “auto”. Note: fitting on sparse input will override the setting of this parameter, using brute force.

  • leaf_size (int) – Leaf size passed to BallTree or KDTree. This can affect the speed of the construction and query, as well as the memory required to store the tree. The optimal value depends on the nature of the problem. Defaults to 30.

  • p (int) – Power parameter for the Minkowski metric. When p = 1, this is equivalent to using manhattan_distance (l1), and euclidean_distance (l2) for p = 2. For arbitrary p, minkowski_distance (l_p) is used. Defaults to 2.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_neighbors”: Integer(2, 12), “weights”: [“uniform”, “distance”], “algorithm”: [“auto”, “ball_tree”, “kd_tree”, “brute”], “leaf_size”: Integer(10, 30), “p”: Integer(1, 5),}

model_family

ModelFamily.K_NEIGHBORS

modifies_features

True

modifies_target

False

name

KNN Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns array of 0’s matching the input number of features as feature_importance is not defined for KNN classifiers.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns array of 0’s matching the input number of features as feature_importance is not defined for KNN classifiers.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.LightGBMClassifier(boosting_type='gbdt', learning_rate=0.1, n_estimators=100, max_depth=0, num_leaves=31, min_child_samples=20, bagging_fraction=0.9, bagging_freq=0, n_jobs=- 1, random_seed=0, **kwargs)[source]

LightGBM Classifier.

Parameters
  • boosting_type (string) – Type of boosting to use. Defaults to “gbdt”. - ‘gbdt’ uses traditional Gradient Boosting Decision Tree - “dart”, uses Dropouts meet Multiple Additive Regression Trees - “goss”, uses Gradient-based One-Side Sampling - “rf”, uses Random Forest

  • learning_rate (float) – Boosting learning rate. Defaults to 0.1.

  • n_estimators (int) – Number of boosted trees to fit. Defaults to 100.

  • max_depth (int) – Maximum tree depth for base learners, <=0 means no limit. Defaults to 0.

  • num_leaves (int) – Maximum tree leaves for base learners. Defaults to 31.

  • min_child_samples (int) – Minimum number of data needed in a child (leaf). Defaults to 20.

  • bagging_fraction (float) – LightGBM will randomly select a subset of features on each iteration (tree) without resampling if this is smaller than 1.0. For example, if set to 0.8, LightGBM will select 80% of features before training each tree. This can be used to speed up training and deal with overfitting. Defaults to 0.9.

  • bagging_freq (int) – Frequency for bagging. 0 means bagging is disabled. k means perform bagging at every k iteration. Every k-th iteration, LightGBM will randomly select bagging_fraction * 100 % of the data to use for the next k iterations. Defaults to 0.

  • n_jobs (int or None) – Number of threads to run in parallel. -1 uses all threads. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “learning_rate”: Real(0.000001, 1), “boosting_type”: [“gbdt”, “dart”, “goss”, “rf”], “n_estimators”: Integer(10, 100), “max_depth”: Integer(0, 10), “num_leaves”: Integer(2, 100), “min_child_samples”: Integer(1, 100), “bagging_fraction”: Real(0.000001, 1), “bagging_freq”: Integer(0, 1),}

model_family

ModelFamily.LIGHTGBM

modifies_features

True

modifies_target

False

name

LightGBM Classifier

SEED_MAX

SEED_BOUNDS.max_bound

SEED_MIN

0

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits LightGBM classifier component to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using the fitted LightGBM classifier.

predict_proba

Make prediction probabilities using the fitted LightGBM classifier.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)[source]

Fits LightGBM classifier component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)[source]

Make predictions using the fitted LightGBM classifier.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.DataFrame

predict_proba(self, X)[source]

Make prediction probabilities using the fitted LightGBM classifier.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted probability values.

Return type

pd.DataFrame

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.LightGBMRegressor(boosting_type='gbdt', learning_rate=0.1, n_estimators=20, max_depth=0, num_leaves=31, min_child_samples=20, bagging_fraction=0.9, bagging_freq=0, n_jobs=- 1, random_seed=0, **kwargs)[source]

LightGBM Regressor.

Parameters
  • boosting_type (string) – Type of boosting to use. Defaults to “gbdt”. - ‘gbdt’ uses traditional Gradient Boosting Decision Tree - “dart”, uses Dropouts meet Multiple Additive Regression Trees - “goss”, uses Gradient-based One-Side Sampling - “rf”, uses Random Forest

  • learning_rate (float) – Boosting learning rate. Defaults to 0.1.

  • n_estimators (int) – Number of boosted trees to fit. Defaults to 100.

  • max_depth (int) – Maximum tree depth for base learners, <=0 means no limit. Defaults to 0.

  • num_leaves (int) – Maximum tree leaves for base learners. Defaults to 31.

  • min_child_samples (int) – Minimum number of data needed in a child (leaf). Defaults to 20.

  • bagging_fraction (float) – LightGBM will randomly select a subset of features on each iteration (tree) without resampling if this is smaller than 1.0. For example, if set to 0.8, LightGBM will select 80% of features before training each tree. This can be used to speed up training and deal with overfitting. Defaults to 0.9.

  • bagging_freq (int) – Frequency for bagging. 0 means bagging is disabled. k means perform bagging at every k iteration. Every k-th iteration, LightGBM will randomly select bagging_fraction * 100 % of the data to use for the next k iterations. Defaults to 0.

  • n_jobs (int or None) – Number of threads to run in parallel. -1 uses all threads. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “learning_rate”: Real(0.000001, 1), “boosting_type”: [“gbdt”, “dart”, “goss”, “rf”], “n_estimators”: Integer(10, 100), “max_depth”: Integer(0, 10), “num_leaves”: Integer(2, 100), “min_child_samples”: Integer(1, 100), “bagging_fraction”: Real(0.000001, 1), “bagging_freq”: Integer(0, 1),}

model_family

ModelFamily.LIGHTGBM

modifies_features

True

modifies_target

False

name

LightGBM Regressor

SEED_MAX

SEED_BOUNDS.max_bound

SEED_MIN

0

supported_problem_types

[ProblemTypes.REGRESSION]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits LightGBM regressor to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using fitted LightGBM regressor.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)[source]

Fits LightGBM regressor to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)[source]

Make predictions using fitted LightGBM regressor.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.LinearRegressor(fit_intercept=True, normalize=False, n_jobs=- 1, random_seed=0, **kwargs)[source]

Linear Regressor.

Parameters
  • fit_intercept (boolean) – Whether to calculate the intercept for this model. If set to False, no intercept will be used in calculations (i.e. data is expected to be centered). Defaults to True.

  • normalize (boolean) – If True, the regressors will be normalized before regression by subtracting the mean and dividing by the l2-norm. This parameter is ignored when fit_intercept is set to False. Defaults to False.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all threads. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “fit_intercept”: [True, False], “normalize”: [True, False]}

model_family

ModelFamily.LINEAR_MODEL

modifies_features

True

modifies_target

False

name

Linear Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance for fitted linear regressor.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance for fitted linear regressor.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.LogisticRegressionClassifier(penalty='l2', C=1.0, multi_class='auto', solver='lbfgs', n_jobs=- 1, random_seed=0, **kwargs)[source]

Logistic Regression Classifier.

Parameters
  • penalty ({"l1", "l2", "elasticnet", "none"}) – The norm used in penalization. Defaults to “l2”.

  • C (float) – Inverse of regularization strength. Must be a positive float. Defaults to 1.0.

  • multi_class ({"auto", "ovr", "multinomial"}) – If the option chosen is “ovr”, then a binary problem is fit for each label. For “multinomial” the loss minimised is the multinomial loss fit across the entire probability distribution, even when the data is binary. “multinomial” is unavailable when solver=”liblinear”. “auto” selects “ovr” if the data is binary, or if solver=”liblinear”, and otherwise selects “multinomial”. Defaults to “auto”.

  • solver ({"newton-cg", "lbfgs", "liblinear", "sag", "saga"}) –

    Algorithm to use in the optimization problem. For small datasets, “liblinear” is a good choice, whereas “sag” and “saga” are faster for large ones. For multiclass problems, only “newton-cg”, “sag”, “saga” and “lbfgs” handle multinomial loss; “liblinear” is limited to one-versus-rest schemes.

    • ”newton-cg”, “lbfgs”, “sag” and “saga” handle L2 or no penalty

    • ”liblinear” and “saga” also handle L1 penalty

    • ”saga” also supports “elasticnet” penalty

    • ”liblinear” does not support setting penalty=’none’

    Defaults to “lbfgs”.

  • n_jobs (int) – Number of parallel threads used to run xgboost. Note that creating thread contention will significantly slow down the algorithm. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “penalty”: [“l2”], “C”: Real(0.01, 10),}

model_family

ModelFamily.LINEAR_MODEL

modifies_features

True

modifies_target

False

name

Logistic Regression Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance for fitted logistic regression classifier.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance for fitted logistic regression classifier.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.MulticlassClassificationPipeline(component_graph, parameters=None, custom_name=None, random_seed=0)[source]

Pipeline subclass for all multiclass classification pipelines.

Parameters
  • component_graph (ComponentGraph, list, dict) – ComponentGraph instance, list of components in order, or dictionary of components. Accepts strings or ComponentBase subclasses in the list. Note that when duplicate components are specified in a list, the duplicate component names will be modified with the component’s index in the list. For example, the component graph [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names [“Imputer”, “One Hot Encoder”, “Imputer_2”, “Logistic Regression Classifier”]

  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • custom_name (str) – Custom name for the pipeline. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Example

>>> pipeline = MulticlassClassificationPipeline(component_graph=["Simple Imputer", "Logistic Regression Classifier"],
...                                             parameters={"Logistic Regression Classifier": {"penalty": "elasticnet",
...                                                                                            "solver": "liblinear"}},
...                                             custom_name="My Multiclass Pipeline")
...
>>> assert pipeline.custom_name == "My Multiclass Pipeline"
>>> assert pipeline.component_graph.component_dict.keys() == {'Simple Imputer', 'Logistic Regression Classifier'}

The pipeline parameters will be chosen from the default parameters for every component, unless specific parameters were passed in as they were above.

>>> assert pipeline.parameters == {
...     'Simple Imputer': {'impute_strategy': 'most_frequent', 'fill_value': None},
...     'Logistic Regression Classifier': {'penalty': 'elasticnet',
...                                        'C': 1.0,
...                                        'n_jobs': -1,
...                                        'multi_class': 'auto',
...                                        'solver': 'liblinear'}}

Attributes

problem_type

ProblemTypes.MULTICLASS

Methods

can_tune_threshold_with_objective

Determine whether the threshold of a binary classification pipeline can be tuned.

classes_

Gets the class names for the pipeline. Will return None before pipeline is fit.

clone

Constructs a new pipeline with the same components, parameters, and random seed.

create_objectives

Create objective instances from a list of strings or objective classes.

custom_name

Custom name of the pipeline.

describe

Outputs pipeline details including component parameters.

feature_importance

Importance associated with each feature. Features dropped by the feature selection are excluded.

fit

Build a classification model. For string and categorical targets, classes are sorted by sorted(set(y)) and then are mapped to values between 0 and n_classes-1.

get_component

Returns component by name.

get_hyperparameter_ranges

Returns hyperparameter ranges from all components as a dictionary.

graph

Generate an image representing the pipeline graph.

graph_feature_importance

Generate a bar graph of the pipeline’s feature importance.

graph_json

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

inverse_transform

Apply component inverse_transform methods to estimator predictions in reverse order.

load

Loads pipeline at file path.

model_family

Returns model family of this pipeline.

name

Name of the pipeline.

new

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

parameters

Parameter dictionary for this pipeline.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves pipeline at file path.

score

Evaluate model performance on objectives.

summary

A short summary of the pipeline structure, describing the list of components used.

transform

Transform the input.

transform_all_but_final

Transforms the data by applying all pre-processing components.

can_tune_threshold_with_objective(self, objective)

Determine whether the threshold of a binary classification pipeline can be tuned.

Parameters

objective (ObjectiveBase) – Primary AutoMLSearch objective.

Returns

True if the pipeline threshold can be tuned.

Return type

bool

property classes_(self)

Gets the class names for the pipeline. Will return None before pipeline is fit.

clone(self)

Constructs a new pipeline with the same components, parameters, and random seed.

Returns

A new instance of this pipeline with identical components, parameters, and random seed.

static create_objectives(objectives)

Create objective instances from a list of strings or objective classes.

property custom_name(self)

Custom name of the pipeline.

describe(self, return_dict=False)

Outputs pipeline details including component parameters.

Parameters

return_dict (bool) – If True, return dictionary of information about pipeline. Defaults to False.

Returns

Dictionary of all component parameters if return_dict is True, else None.

Return type

dict

property feature_importance(self)

Importance associated with each feature. Features dropped by the feature selection are excluded.

Returns

Feature names and their corresponding importance

Return type

pd.DataFrame

fit(self, X, y)

Build a classification model. For string and categorical targets, classes are sorted by sorted(set(y)) and then are mapped to values between 0 and n_classes-1.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, np.ndarray) – The target training labels of length [n_samples]

Returns

self

get_component(self, name)

Returns component by name.

Parameters

name (str) – Name of component.

Returns

Component to return

Return type

Component

get_hyperparameter_ranges(self, custom_hyperparameters)

Returns hyperparameter ranges from all components as a dictionary.

Parameters

custom_hyperparameters (dict) – Custom hyperparameters for the pipeline.

Returns

Dictionary of hyperparameter ranges for each component in the pipeline.

Return type

dict

graph(self, filepath=None)

Generate an image representing the pipeline graph.

Parameters

filepath (str, optional) – Path to where the graph should be saved. If set to None (as by default), the graph will not be saved.

Returns

Graph object that can be directly displayed in Jupyter notebooks.

Return type

graphviz.Digraph

Raises
  • RuntimeError – If graphviz is not installed.

  • ValueError – If path is not writeable.

graph_feature_importance(self, importance_threshold=0)

Generate a bar graph of the pipeline’s feature importance.

Parameters

importance_threshold (float, optional) – If provided, graph features with a permutation importance whose absolute value is larger than importance_threshold. Defaults to zero.

Returns

A bar graph showing features and their corresponding importance.

Return type

plotly.Figure

Raises

ValueError – If importance threshold is not valid.

graph_json(self)

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

x_edges specifies from which component feature data is being passed. y_edges specifies from which component target data is being passed. This can be used to build graphs across a variety of visualization tools. Template: {“Nodes”: {“component_name”: {“Name”: class_name, “Parameters”: parameters_attributes}, …}}, “x_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …], “y_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …]}

Returns

A serialized JSON representation of a DAG structure.

Return type

dag_json (str)

inverse_transform(self, y)

Apply component inverse_transform methods to estimator predictions in reverse order.

Components that implement inverse_transform are PolynomialDetrender, LogTransformer, LabelEncoder (tbd).

Parameters

y (pd.Series) – Final component features.

Returns

The inverse transform of the target.

Return type

pd.Series

static load(file_path)

Loads pipeline at file path.

Parameters

file_path (str) – Location to load file.

Returns

PipelineBase object

property model_family(self)

Returns model family of this pipeline.

property name(self)

Name of the pipeline.

new(self, parameters, random_seed=0)

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

Parameters
  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Returns

A new instance of this pipeline with identical components.

property parameters(self)

Parameter dictionary for this pipeline.

Returns

Dictionary of all component parameters.

Return type

dict

predict(self, X, objective=None, X_train=None, y_train=None)

Make predictions using selected features.

Note: we cast y as ints first to address boolean values that may be returned from calculating predictions which we would not be able to otherwise transform if we originally had integer targets.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features].

  • objective (Object or string) – The objective to use to make predictions.

  • X_train (pd.DataFrame) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series) – Training labels. Ignored. Only used for time series.

Returns

Estimated labels.

Return type

pd.Series

predict_proba(self, X, X_train=None, y_train=None)

Make probability estimates for labels.

Parameters
  • X (pd.DataFrame or np.ndarray) – Data of shape [n_samples, n_features]

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Probability estimates

Return type

pd.DataFrame

Raises

ValueError – If final component is not an estimator.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves pipeline at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

score(self, X, y, objectives, X_train=None, y_train=None)

Evaluate model performance on objectives.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features]

  • y (pd.Series) – True labels of length [n_samples]

  • objectives (list) – List of objectives to score

  • X_train (pd.DataFrame) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series) – Training labels. Ignored. Only used for time series.

Returns

Ordered dictionary of objective scores.

Return type

dict

property summary(self)

A short summary of the pipeline structure, describing the list of components used.

Example: Logistic Regression Classifier w/ Simple Imputer + One Hot Encoder

Returns

A string describing the pipeline structure.

transform(self, X, y=None)

Transform the input.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • y (pd.Series) – The target data of length [n_samples]. Defaults to None.

Returns

Transformed output.

Return type

pd.DataFrame

transform_all_but_final(self, X, y=None, X_train=None, y_train=None)

Transforms the data by applying all pre-processing components.

Parameters
  • X (pd.DataFrame) – Input data to the pipeline to transform.

  • y (pd.Series or None) – Targets corresponding to X. Optional.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Only used for time series.

Returns

New transformed features.

Return type

pd.DataFrame

class evalml.pipelines.OneHotEncoder(top_n=10, features_to_encode=None, categories=None, drop='if_binary', handle_unknown='ignore', handle_missing='error', random_seed=0, **kwargs)[source]

A transformer that encodes categorical features in a one-hot numeric array.

Parameters
  • top_n (int) – Number of categories per column to encode. If None, all categories will be encoded. Otherwise, the n most frequent will be encoded and all others will be dropped. Defaults to 10.

  • features_to_encode (list[str]) – List of columns to encode. All other columns will remain untouched. If None, all appropriate columns will be encoded. Defaults to None.

  • categories (list) – A two dimensional list of categories, where categories[i] is a list of the categories for the column at index i. This can also be None, or “auto” if top_n is not None. Defaults to None.

  • drop (string, list) – Method (“first” or “if_binary”) to use to drop one category per feature. Can also be a list specifying which categories to drop for each feature. Defaults to ‘if_binary’.

  • handle_unknown (string) – Whether to ignore or error for unknown categories for a feature encountered during fit or transform. If either top_n or categories is used to limit the number of categories per column, this must be “ignore”. Defaults to “ignore”.

  • handle_missing (string) – Options for how to handle missing (NaN) values encountered during fit or transform. If this is set to “as_category” and NaN values are within the n most frequent, “nan” values will be encoded as their own column. If this is set to “error”, any missing values encountered will raise an error. Defaults to “error”.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{}

modifies_features

True

modifies_target

False

name

One Hot Encoder

training_only

False

Methods

categories

Returns a list of the unique categories to be encoded for the particular feature, in order.

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits the one-hot encoder component.

fit_transform

Fits on X and transforms X.

get_feature_names

Return feature names for the categorical features after fitting.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

One-hot encode the input data.

categories(self, feature_name)[source]

Returns a list of the unique categories to be encoded for the particular feature, in order.

Parameters

feature_name (str) – The name of any feature provided to one-hot encoder during fit.

Returns

The unique categories, in the same dtype as they were provided during fit.

Return type

np.ndarray

Raises

ValueError – If feature was not provided to one-hot encoder as a training feature.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)[source]

Fits the one-hot encoder component.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

Raises

ValueError – If encoding a column failed.

fit_transform(self, X, y=None)

Fits on X and transforms X.

Parameters
  • X (pd.DataFrame) – Data to fit and transform.

  • y (pd.Series) – Target data.

Returns

Transformed X.

Return type

pd.DataFrame

Raises

MethodPropertyNotFoundError – If transformer does not have a transform method or a component_obj that implements transform.

get_feature_names(self)[source]

Return feature names for the categorical features after fitting.

Feature names are formatted as {column name}_{category name}. In the event of a duplicate name, an integer will be added at the end of the feature name to distinguish it.

For example, consider a dataframe with a column called “A” and category “x_y” and another column called “A_x” with “y”. In this example, the feature names would be “A_x_y” and “A_x_y_1”.

Returns

The feature names after encoding, provided in the same order as input_features.

Return type

np.ndarray

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)[source]

One-hot encode the input data.

Parameters
  • X (pd.DataFrame) – Features to one-hot encode.

  • y (pd.Series) – Ignored.

Returns

Transformed data, where each categorical feature has been encoded into numerical columns using one-hot encoding.

Return type

pd.DataFrame

class evalml.pipelines.PerColumnImputer(impute_strategies=None, default_impute_strategy='most_frequent', impute_all=True, random_seed=0, **kwargs)[source]

Imputes missing data according to a specified imputation strategy per column.

Parameters
  • impute_strategies (dict) – Column and {“impute_strategy”: strategy, “fill_value”:value} pairings. Valid values for impute strategy include “mean”, “median”, “most_frequent”, “constant” for numerical data, and “most_frequent”, “constant” for object data types. Defaults to None, which uses “most_frequent” for all columns. When impute_strategy == “constant”, fill_value is used to replace missing data. When None, uses 0 when imputing numerical data and “missing_value” for strings or object data types.

  • default_impute_strategy (str) – Impute strategy to fall back on when none is provided for a certain column. Valid values include “mean”, “median”, “most_frequent”, “constant” for numerical data, and “most_frequent”, “constant” for object data types. Defaults to “most_frequent”.

  • impute_all (bool) – Whether or not to impute all columns or just the columns that are specified in impute_strategies. If True, columns will be imputed using the strategy in the impute_strategies dictionary if specified or using the default_impute_strategy. If False, only columns specified as keys in the impute_strategies dictionary are imputed. If False and impute_strategies is None, no columns will be imputed. Defaults to True.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{}

modifies_features

True

modifies_target

False

name

Per Column Imputer

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits imputers on input data.

fit_transform

Fits on X and transforms X.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Transforms input data by imputing missing values.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)[source]

Fits imputers on input data.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features] to fit.

  • y (pd.Series, optional) – The target training data of length [n_samples]. Ignored.

Returns

self

fit_transform(self, X, y=None)

Fits on X and transforms X.

Parameters
  • X (pd.DataFrame) – Data to fit and transform.

  • y (pd.Series) – Target data.

Returns

Transformed X.

Return type

pd.DataFrame

Raises

MethodPropertyNotFoundError – If transformer does not have a transform method or a component_obj that implements transform.

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)[source]

Transforms input data by imputing missing values.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features] to transform.

  • y (pd.Series, optional) – The target training data of length [n_samples]. Ignored.

Returns

Transformed X

Return type

pd.DataFrame

class evalml.pipelines.PipelineBase(component_graph, parameters=None, custom_name=None, random_seed=0)[source]

Machine learning pipeline.

Parameters
  • component_graph (ComponentGraph, list, dict) – ComponentGraph instance, list of components in order, or dictionary of components. Accepts strings or ComponentBase subclasses in the list. Note that when duplicate components are specified in a list, the duplicate component names will be modified with the component’s index in the list. For example, the component graph [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names [“Imputer”, “One Hot Encoder”, “Imputer_2”, “Logistic Regression Classifier”].

  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • custom_name (str) – Custom name for the pipeline. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

problem_type

None

Methods

can_tune_threshold_with_objective

Determine whether the threshold of a binary classification pipeline can be tuned.

clone

Constructs a new pipeline with the same components, parameters, and random seed.

create_objectives

Create objective instances from a list of strings or objective classes.

custom_name

Custom name of the pipeline.

describe

Outputs pipeline details including component parameters.

feature_importance

Importance associated with each feature. Features dropped by the feature selection are excluded.

fit

Build a model.

get_component

Returns component by name.

get_hyperparameter_ranges

Returns hyperparameter ranges from all components as a dictionary.

graph

Generate an image representing the pipeline graph.

graph_feature_importance

Generate a bar graph of the pipeline’s feature importance.

graph_json

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

inverse_transform

Apply component inverse_transform methods to estimator predictions in reverse order.

load

Loads pipeline at file path.

model_family

Returns model family of this pipeline.

name

Name of the pipeline.

new

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

parameters

Parameter dictionary for this pipeline.

predict

Make predictions using selected features.

save

Saves pipeline at file path.

score

Evaluate model performance on current and additional objectives.

summary

A short summary of the pipeline structure, describing the list of components used.

transform

Transform the input.

transform_all_but_final

Transforms the data by applying all pre-processing components.

can_tune_threshold_with_objective(self, objective)[source]

Determine whether the threshold of a binary classification pipeline can be tuned.

Parameters

objective (ObjectiveBase) – Primary AutoMLSearch objective.

Returns

True if the pipeline threshold can be tuned.

Return type

bool

clone(self)[source]

Constructs a new pipeline with the same components, parameters, and random seed.

Returns

A new instance of this pipeline with identical components, parameters, and random seed.

static create_objectives(objectives)[source]

Create objective instances from a list of strings or objective classes.

property custom_name(self)

Custom name of the pipeline.

describe(self, return_dict=False)[source]

Outputs pipeline details including component parameters.

Parameters

return_dict (bool) – If True, return dictionary of information about pipeline. Defaults to False.

Returns

Dictionary of all component parameters if return_dict is True, else None.

Return type

dict

property feature_importance(self)

Importance associated with each feature. Features dropped by the feature selection are excluded.

Returns

Feature names and their corresponding importance

Return type

pd.DataFrame

abstract fit(self, X, y)[source]

Build a model.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, np.ndarray) – The target training data of length [n_samples].

Returns

self

get_component(self, name)[source]

Returns component by name.

Parameters

name (str) – Name of component.

Returns

Component to return

Return type

Component

get_hyperparameter_ranges(self, custom_hyperparameters)[source]

Returns hyperparameter ranges from all components as a dictionary.

Parameters

custom_hyperparameters (dict) – Custom hyperparameters for the pipeline.

Returns

Dictionary of hyperparameter ranges for each component in the pipeline.

Return type

dict

graph(self, filepath=None)[source]

Generate an image representing the pipeline graph.

Parameters

filepath (str, optional) – Path to where the graph should be saved. If set to None (as by default), the graph will not be saved.

Returns

Graph object that can be directly displayed in Jupyter notebooks.

Return type

graphviz.Digraph

Raises
  • RuntimeError – If graphviz is not installed.

  • ValueError – If path is not writeable.

graph_feature_importance(self, importance_threshold=0)[source]

Generate a bar graph of the pipeline’s feature importance.

Parameters

importance_threshold (float, optional) – If provided, graph features with a permutation importance whose absolute value is larger than importance_threshold. Defaults to zero.

Returns

A bar graph showing features and their corresponding importance.

Return type

plotly.Figure

Raises

ValueError – If importance threshold is not valid.

graph_json(self)[source]

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

x_edges specifies from which component feature data is being passed. y_edges specifies from which component target data is being passed. This can be used to build graphs across a variety of visualization tools. Template: {“Nodes”: {“component_name”: {“Name”: class_name, “Parameters”: parameters_attributes}, …}}, “x_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …], “y_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …]}

Returns

A serialized JSON representation of a DAG structure.

Return type

dag_json (str)

inverse_transform(self, y)[source]

Apply component inverse_transform methods to estimator predictions in reverse order.

Components that implement inverse_transform are PolynomialDetrender, LogTransformer, LabelEncoder (tbd).

Parameters

y (pd.Series) – Final component features.

Returns

The inverse transform of the target.

Return type

pd.Series

static load(file_path)[source]

Loads pipeline at file path.

Parameters

file_path (str) – Location to load file.

Returns

PipelineBase object

property model_family(self)

Returns model family of this pipeline.

property name(self)

Name of the pipeline.

new(self, parameters, random_seed=0)[source]

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

Parameters
  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Returns

A new instance of this pipeline with identical components.

property parameters(self)

Parameter dictionary for this pipeline.

Returns

Dictionary of all component parameters.

Return type

dict

predict(self, X, objective=None, X_train=None, y_train=None)[source]

Make predictions using selected features.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • objective (Object or string) – The objective to use to make predictions.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Predicted values.

Return type

pd.Series

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)[source]

Saves pipeline at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

abstract score(self, X, y, objectives, X_train=None, y_train=None)[source]

Evaluate model performance on current and additional objectives.

Parameters
  • X (pd.DataFrame or np.ndarray) – Data of shape [n_samples, n_features].

  • y (pd.Series, np.ndarray) – True labels of length [n_samples].

  • objectives (list) – Non-empty list of objectives to score on.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Ordered dictionary of objective scores.

Return type

dict

property summary(self)

A short summary of the pipeline structure, describing the list of components used.

Example: Logistic Regression Classifier w/ Simple Imputer + One Hot Encoder

Returns

A string describing the pipeline structure.

transform(self, X, y=None)[source]

Transform the input.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • y (pd.Series) – The target data of length [n_samples]. Defaults to None.

Returns

Transformed output.

Return type

pd.DataFrame

transform_all_but_final(self, X, y=None, X_train=None, y_train=None)[source]

Transforms the data by applying all pre-processing components.

Parameters
  • X (pd.DataFrame) – Input data to the pipeline to transform.

  • y (pd.Series or None) – Targets corresponding to X. Optional.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Only used for time series.

Returns

New transformed features.

Return type

pd.DataFrame

class evalml.pipelines.ProphetRegressor(time_index=None, changepoint_prior_scale=0.05, seasonality_prior_scale=10, holidays_prior_scale=10, seasonality_mode='additive', random_seed=0, stan_backend='CMDSTANPY', **kwargs)[source]

Prophet is a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. It works best with time series that have strong seasonal effects and several seasons of historical data. Prophet is robust to missing data and shifts in the trend, and typically handles outliers well.

More information here: https://facebook.github.io/prophet/

Attributes

hyperparameter_ranges

{ “changepoint_prior_scale”: Real(0.001, 0.5), “seasonality_prior_scale”: Real(0.01, 10), “holidays_prior_scale”: Real(0.01, 10), “seasonality_mode”: [“additive”, “multiplicative”],}

model_family

ModelFamily.PROPHET

modifies_features

True

modifies_target

False

name

Prophet Regressor

supported_problem_types

[ProblemTypes.TIME_SERIES_REGRESSION]

training_only

False

Methods

build_prophet_df

Build the Prophet data to pass fit and predict on.

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns array of 0’s with len(1) as feature_importance is not defined for Prophet regressor.

fit

Fits Prophet regressor component to data.

get_params

Get parameters for the Prophet regressor.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using fitted Prophet regressor.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

static build_prophet_df(X, y=None, time_index='ds')[source]

Build the Prophet data to pass fit and predict on.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns array of 0’s with len(1) as feature_importance is not defined for Prophet regressor.

fit(self, X, y=None)[source]

Fits Prophet regressor component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series) – The target training data of length [n_samples].

Returns

self

get_params(self)[source]

Get parameters for the Prophet regressor.

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X, y=None)[source]

Make predictions using fitted Prophet regressor.

Parameters
  • X (pd.DataFrame) – Data of shape [n_samples, n_features].

  • y (pd.Series) – Target data. Ignored.

Returns

Predicted values.

Return type

pd.Series

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.RandomForestClassifier(n_estimators=100, max_depth=6, n_jobs=- 1, random_seed=0, **kwargs)[source]

Random Forest Classifier.

Parameters
  • n_estimators (float) – The number of trees in the forest. Defaults to 100.

  • max_depth (int) – Maximum tree depth for base learners. Defaults to 6.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_estimators”: Integer(10, 1000), “max_depth”: Integer(1, 10),}

model_family

ModelFamily.RANDOM_FOREST

modifies_features

True

modifies_target

False

name

Random Forest Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.RandomForestRegressor(n_estimators=100, max_depth=6, n_jobs=- 1, random_seed=0, **kwargs)[source]

Random Forest Regressor.

Parameters
  • n_estimators (float) – The number of trees in the forest. Defaults to 100.

  • max_depth (int) – Maximum tree depth for base learners. Defaults to 6.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “n_estimators”: Integer(10, 1000), “max_depth”: Integer(1, 32),}

model_family

ModelFamily.RANDOM_FOREST

modifies_features

True

modifies_target

False

name

Random Forest Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Returns importance associated with each feature.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Returns importance associated with each feature.

Returns

Importance associated with each feature.

Return type

np.ndarray

Raises

MethodPropertyNotFoundError – If estimator does not have a feature_importance method or a component_obj that implements feature_importance.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.RegressionPipeline(component_graph, parameters=None, custom_name=None, random_seed=0)[source]

Pipeline subclass for all regression pipelines.

Parameters
  • component_graph (ComponentGraph, list, dict) – ComponentGraph instance, list of components in order, or dictionary of components. Accepts strings or ComponentBase subclasses in the list. Note that when duplicate components are specified in a list, the duplicate component names will be modified with the component’s index in the list. For example, the component graph [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names [“Imputer”, “One Hot Encoder”, “Imputer_2”, “Logistic Regression Classifier”]

  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • custom_name (str) – Custom name for the pipeline. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Example

>>> pipeline = RegressionPipeline(component_graph=["Simple Imputer", "Linear Regressor"],
...                               parameters={"Linear Regressor": {"normalize": True}},
...                               custom_name="My Regression Pipeline")
...
>>> assert pipeline.custom_name == "My Regression Pipeline"
>>> assert pipeline.component_graph.component_dict.keys() == {'Simple Imputer', 'Linear Regressor'}

The pipeline parameters will be chosen from the default parameters for every component, unless specific parameters were passed in as they were above.

>>> assert pipeline.parameters == {
...     'Simple Imputer': {'impute_strategy': 'most_frequent', 'fill_value': None},
...     'Linear Regressor': {'fit_intercept': True, 'normalize': True, 'n_jobs': -1}}

Attributes

problem_type

ProblemTypes.REGRESSION

Methods

can_tune_threshold_with_objective

Determine whether the threshold of a binary classification pipeline can be tuned.

clone

Constructs a new pipeline with the same components, parameters, and random seed.

create_objectives

Create objective instances from a list of strings or objective classes.

custom_name

Custom name of the pipeline.

describe

Outputs pipeline details including component parameters.

feature_importance

Importance associated with each feature. Features dropped by the feature selection are excluded.

fit

Build a regression model.

get_component

Returns component by name.

get_hyperparameter_ranges

Returns hyperparameter ranges from all components as a dictionary.

graph

Generate an image representing the pipeline graph.

graph_feature_importance

Generate a bar graph of the pipeline’s feature importance.

graph_json

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

inverse_transform

Apply component inverse_transform methods to estimator predictions in reverse order.

load

Loads pipeline at file path.

model_family

Returns model family of this pipeline.

name

Name of the pipeline.

new

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

parameters

Parameter dictionary for this pipeline.

predict

Make predictions using selected features.

save

Saves pipeline at file path.

score

Evaluate model performance on current and additional objectives.

summary

A short summary of the pipeline structure, describing the list of components used.

transform

Transform the input.

transform_all_but_final

Transforms the data by applying all pre-processing components.

can_tune_threshold_with_objective(self, objective)

Determine whether the threshold of a binary classification pipeline can be tuned.

Parameters

objective (ObjectiveBase) – Primary AutoMLSearch objective.

Returns

True if the pipeline threshold can be tuned.

Return type

bool

clone(self)

Constructs a new pipeline with the same components, parameters, and random seed.

Returns

A new instance of this pipeline with identical components, parameters, and random seed.

static create_objectives(objectives)

Create objective instances from a list of strings or objective classes.

property custom_name(self)

Custom name of the pipeline.

describe(self, return_dict=False)

Outputs pipeline details including component parameters.

Parameters

return_dict (bool) – If True, return dictionary of information about pipeline. Defaults to False.

Returns

Dictionary of all component parameters if return_dict is True, else None.

Return type

dict

property feature_importance(self)

Importance associated with each feature. Features dropped by the feature selection are excluded.

Returns

Feature names and their corresponding importance

Return type

pd.DataFrame

fit(self, X, y)[source]

Build a regression model.

Parameters
  • X (pd.DataFrame or np.ndarray) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, np.ndarray) – The target training data of length [n_samples]

Returns

self

Raises

ValueError – If the target is not numeric.

get_component(self, name)

Returns component by name.

Parameters

name (str) – Name of component.

Returns

Component to return

Return type

Component

get_hyperparameter_ranges(self, custom_hyperparameters)

Returns hyperparameter ranges from all components as a dictionary.

Parameters

custom_hyperparameters (dict) – Custom hyperparameters for the pipeline.

Returns

Dictionary of hyperparameter ranges for each component in the pipeline.

Return type

dict

graph(self, filepath=None)

Generate an image representing the pipeline graph.

Parameters

filepath (str, optional) – Path to where the graph should be saved. If set to None (as by default), the graph will not be saved.

Returns

Graph object that can be directly displayed in Jupyter notebooks.

Return type

graphviz.Digraph

Raises
  • RuntimeError – If graphviz is not installed.

  • ValueError – If path is not writeable.

graph_feature_importance(self, importance_threshold=0)

Generate a bar graph of the pipeline’s feature importance.

Parameters

importance_threshold (float, optional) – If provided, graph features with a permutation importance whose absolute value is larger than importance_threshold. Defaults to zero.

Returns

A bar graph showing features and their corresponding importance.

Return type

plotly.Figure

Raises

ValueError – If importance threshold is not valid.

graph_json(self)

Generates a JSON with nodes consisting of the component names and parameters, and edges detailing component relationships.

x_edges specifies from which component feature data is being passed. y_edges specifies from which component target data is being passed. This can be used to build graphs across a variety of visualization tools. Template: {“Nodes”: {“component_name”: {“Name”: class_name, “Parameters”: parameters_attributes}, …}}, “x_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …], “y_edges”: [[from_component_name, to_component_name], [from_component_name, to_component_name], …]}

Returns

A serialized JSON representation of a DAG structure.

Return type

dag_json (str)

inverse_transform(self, y)

Apply component inverse_transform methods to estimator predictions in reverse order.

Components that implement inverse_transform are PolynomialDetrender, LogTransformer, LabelEncoder (tbd).

Parameters

y (pd.Series) – Final component features.

Returns

The inverse transform of the target.

Return type

pd.Series

static load(file_path)

Loads pipeline at file path.

Parameters

file_path (str) – Location to load file.

Returns

PipelineBase object

property model_family(self)

Returns model family of this pipeline.

property name(self)

Name of the pipeline.

new(self, parameters, random_seed=0)

Constructs a new instance of the pipeline with the same component graph but with a different set of parameters. Not to be confused with python’s __new__ method.

Parameters
  • parameters (dict) – Dictionary with component names as keys and dictionary of that component’s parameters as values. An empty dictionary or None implies using all default values for component parameters. Defaults to None.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Returns

A new instance of this pipeline with identical components.

property parameters(self)

Parameter dictionary for this pipeline.

Returns

Dictionary of all component parameters.

Return type

dict

predict(self, X, objective=None, X_train=None, y_train=None)[source]

Make predictions using selected features.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • objective (Object or string) – The objective to use to make predictions.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Predicted values.

Return type

pd.Series

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves pipeline at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

score(self, X, y, objectives, X_train=None, y_train=None)[source]

Evaluate model performance on current and additional objectives.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features]

  • y (pd.Series, or np.ndarray) – True values of length [n_samples]

  • objectives (list) – Non-empty list of objectives to score on

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Ignored. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Ignored. Only used for time series.

Returns

Ordered dictionary of objective scores.

Return type

dict

property summary(self)

A short summary of the pipeline structure, describing the list of components used.

Example: Logistic Regression Classifier w/ Simple Imputer + One Hot Encoder

Returns

A string describing the pipeline structure.

transform(self, X, y=None)

Transform the input.

Parameters
  • X (pd.DataFrame, or np.ndarray) – Data of shape [n_samples, n_features].

  • y (pd.Series) – The target data of length [n_samples]. Defaults to None.

Returns

Transformed output.

Return type

pd.DataFrame

transform_all_but_final(self, X, y=None, X_train=None, y_train=None)

Transforms the data by applying all pre-processing components.

Parameters
  • X (pd.DataFrame) – Input data to the pipeline to transform.

  • y (pd.Series or None) – Targets corresponding to X. Optional.

  • X_train (pd.DataFrame or np.ndarray or None) – Training data. Only used for time series.

  • y_train (pd.Series or None) – Training labels. Only used for time series.

Returns

New transformed features.

Return type

pd.DataFrame

class evalml.pipelines.RFClassifierSelectFromModel(number_features=None, n_estimators=10, max_depth=None, percent_features=0.5, threshold='median', n_jobs=- 1, random_seed=0, **kwargs)[source]

Selects top features based on importance weights using a Random Forest classifier.

Parameters
  • number_features (int) – The maximum number of features to select. If both percent_features and number_features are specified, take the greater number of features. Defaults to 0.5. Defaults to None.

  • n_estimators (float) – The number of trees in the forest. Defaults to 100.

  • max_depth (int) – Maximum tree depth for base learners. Defaults to 6.

  • percent_features (float) – Percentage of features to use. If both percent_features and number_features are specified, take the greater number of features. Defaults to 0.5.

  • threshold (string or float) – The threshold value to use for feature selection. Features whose importance is greater or equal are kept while the others are discarded. If “median”, then the threshold value is the median of the feature importances. A scaling factor (e.g., “1.25*mean”) may also be used. Defaults to -np.inf.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “percent_features”: Real(0.01, 1), “threshold”: [“mean”, “median”],}

modifies_features

True

modifies_target

False

name

RF Classifier Select From Model

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits component to data.

fit_transform

Fit and transform data using the feature selector.

get_names

Get names of selected features.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Transforms input data by selecting features. If the component_obj does not have a transform method, will raise an MethodPropertyNotFoundError exception.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)

Fits component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, optional) – The target training data of length [n_samples]

Returns

self

Raises

MethodPropertyNotFoundError – If component does not have a fit method or a component_obj that implements fit.

fit_transform(self, X, y=None)

Fit and transform data using the feature selector.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

Transformed data.

Return type

pd.DataFrame

get_names(self)

Get names of selected features.

Returns

List of the names of features selected.

Return type

list[str]

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)

Transforms input data by selecting features. If the component_obj does not have a transform method, will raise an MethodPropertyNotFoundError exception.

Parameters
  • X (pd.DataFrame) – Data to transform.

  • y (pd.Series, optional) – Target data. Ignored.

Returns

Transformed X

Return type

pd.DataFrame

Raises

MethodPropertyNotFoundError – If feature selector does not have a transform method or a component_obj that implements transform

class evalml.pipelines.RFRegressorSelectFromModel(number_features=None, n_estimators=10, max_depth=None, percent_features=0.5, threshold='median', n_jobs=- 1, random_seed=0, **kwargs)[source]

Selects top features based on importance weights using a Random Forest regressor.

Parameters
  • number_features (int) – The maximum number of features to select. If both percent_features and number_features are specified, take the greater number of features. Defaults to 0.5. Defaults to None.

  • n_estimators (float) – The number of trees in the forest. Defaults to 100.

  • max_depth (int) – Maximum tree depth for base learners. Defaults to 6.

  • percent_features (float) – Percentage of features to use. If both percent_features and number_features are specified, take the greater number of features. Defaults to 0.5.

  • threshold (string or float) – The threshold value to use for feature selection. Features whose importance is greater or equal are kept while the others are discarded. If “median”, then the threshold value is the median of the feature importances. A scaling factor (e.g., “1.25*mean”) may also be used. Defaults to -np.inf.

  • n_jobs (int or None) – Number of jobs to run in parallel. -1 uses all processes. Defaults to -1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “percent_features”: Real(0.01, 1), “threshold”: [“mean”, “median”],}

modifies_features

True

modifies_target

False

name

RF Regressor Select From Model

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits component to data.

fit_transform

Fit and transform data using the feature selector.

get_names

Get names of selected features.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Transforms input data by selecting features. If the component_obj does not have a transform method, will raise an MethodPropertyNotFoundError exception.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)

Fits component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, optional) – The target training data of length [n_samples]

Returns

self

Raises

MethodPropertyNotFoundError – If component does not have a fit method or a component_obj that implements fit.

fit_transform(self, X, y=None)

Fit and transform data using the feature selector.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

Transformed data.

Return type

pd.DataFrame

get_names(self)

Get names of selected features.

Returns

List of the names of features selected.

Return type

list[str]

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)

Transforms input data by selecting features. If the component_obj does not have a transform method, will raise an MethodPropertyNotFoundError exception.

Parameters
  • X (pd.DataFrame) – Data to transform.

  • y (pd.Series, optional) – Target data. Ignored.

Returns

Transformed X

Return type

pd.DataFrame

Raises

MethodPropertyNotFoundError – If feature selector does not have a transform method or a component_obj that implements transform

class evalml.pipelines.SimpleImputer(impute_strategy='most_frequent', fill_value=None, random_seed=0, **kwargs)[source]

Imputes missing data according to a specified imputation strategy.

Parameters
  • impute_strategy (string) – Impute strategy to use. Valid values include “mean”, “median”, “most_frequent”, “constant” for numerical data, and “most_frequent”, “constant” for object data types.

  • fill_value (string) – When impute_strategy == “constant”, fill_value is used to replace missing data. Defaults to 0 when imputing numerical data and “missing_value” for strings or object data types.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “impute_strategy”: [“mean”, “median”, “most_frequent”]}

modifies_features

True

modifies_target

False

name

Simple Imputer

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits imputer to data. ‘None’ values are converted to np.nan before imputation and are treated as the same.

fit_transform

Fits on X and transforms X.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Transforms input by imputing missing values. ‘None’ and np.nan values are treated as the same.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)[source]

Fits imputer to data. ‘None’ values are converted to np.nan before imputation and are treated as the same.

Parameters
  • X (pd.DataFrame or np.ndarray) – the input training data of shape [n_samples, n_features]

  • y (pd.Series, optional) – the target training data of length [n_samples]

Returns

self

fit_transform(self, X, y=None)[source]

Fits on X and transforms X.

Parameters
  • X (pd.DataFrame) – Data to fit and transform

  • y (pd.Series, optional) – Target data.

Returns

Transformed X

Return type

pd.DataFrame

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)[source]

Transforms input by imputing missing values. ‘None’ and np.nan values are treated as the same.

Parameters
  • X (pd.DataFrame) – Data to transform.

  • y (pd.Series, optional) – Ignored.

Returns

Transformed X

Return type

pd.DataFrame

class evalml.pipelines.StackedEnsembleClassifier(final_estimator=None, n_jobs=- 1, random_seed=0, **kwargs)[source]

Stacked Ensemble Classifier.

Parameters
  • final_estimator (Estimator or subclass) – The classifier used to combine the base estimators. If None, uses ElasticNetClassifier.

  • n_jobs (int or None) – Integer describing level of parallelism used for pipelines. None and 1 are equivalent. If set to -1, all CPUs are used. For n_jobs below -1, (n_cpus + 1 + n_jobs) are used. Defaults to -1. - Note: there could be some multi-process errors thrown for values of n_jobs != 1. If this is the case, please use n_jobs = 1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Example

>>> from evalml.pipelines.component_graph import ComponentGraph
>>> from evalml.pipelines.components.estimators.classifiers.decision_tree_classifier import DecisionTreeClassifier
>>> from evalml.pipelines.components.estimators.classifiers.elasticnet_classifier import ElasticNetClassifier
...
>>> component_graph = {
...     "Decision Tree": [DecisionTreeClassifier(random_seed=3), "X", "y"],
...     "Decision Tree B": [DecisionTreeClassifier(random_seed=4), "X", "y"],
...     "Stacked Ensemble": [
...         StackedEnsembleClassifier(n_jobs=1, final_estimator=DecisionTreeClassifier()),
...         "Decision Tree.x",
...         "Decision Tree B.x",
...         "y",
...     ],
... }
...
>>> cg = ComponentGraph(component_graph)
>>> assert cg.default_parameters == {
...     'Decision Tree Classifier': {'criterion': 'gini',
...                                  'max_features': 'auto',
...                                  'max_depth': 6,
...                                  'min_samples_split': 2,
...                                  'min_weight_fraction_leaf': 0.0},
...     'Stacked Ensemble Classifier': {'final_estimator': ElasticNetClassifier,
...                                     'n_jobs': -1}}

Attributes

hyperparameter_ranges

{}

model_family

ModelFamily.ENSEMBLE

modifies_features

True

modifies_target

False

name

Stacked Ensemble Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for stacked ensemble classes.

describe

Describe a component and its parameters.

feature_importance

Not implemented for StackedEnsembleClassifier and StackedEnsembleRegressor.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for stacked ensemble classes.

Returns

default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Not implemented for StackedEnsembleClassifier and StackedEnsembleRegressor.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.StackedEnsembleRegressor(final_estimator=None, n_jobs=- 1, random_seed=0, **kwargs)[source]

Stacked Ensemble Regressor.

Parameters
  • final_estimator (Estimator or subclass) – The regressor used to combine the base estimators. If None, uses ElasticNetRegressor.

  • n_jobs (int or None) – Integer describing level of parallelism used for pipelines. None and 1 are equivalent. If set to -1, all CPUs are used. For n_jobs greater than -1, (n_cpus + 1 + n_jobs) are used. Defaults to -1. - Note: there could be some multi-process errors thrown for values of n_jobs != 1. If this is the case, please use n_jobs = 1.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Example

>>> from evalml.pipelines.component_graph import ComponentGraph
>>> from evalml.pipelines.components.estimators.regressors.rf_regressor import RandomForestRegressor
>>> from evalml.pipelines.components.estimators.regressors.elasticnet_regressor import ElasticNetRegressor
...
>>> component_graph = {
...     "Random Forest": [RandomForestRegressor(random_seed=3), "X", "y"],
...     "Random Forest B": [RandomForestRegressor(random_seed=4), "X", "y"],
...     "Stacked Ensemble": [
...         StackedEnsembleRegressor(n_jobs=1, final_estimator=RandomForestRegressor()),
...         "Random Forest.x",
...         "Random Forest B.x",
...         "y",
...     ],
... }
...
>>> cg = ComponentGraph(component_graph)
>>> assert cg.default_parameters == {
...     'Random Forest Regressor': {'n_estimators': 100,
...                                 'max_depth': 6,
...                                 'n_jobs': -1},
...     'Stacked Ensemble Regressor': {'final_estimator': ElasticNetRegressor,
...                                    'n_jobs': -1}}

Attributes

hyperparameter_ranges

{}

model_family

ModelFamily.ENSEMBLE

modifies_features

True

modifies_target

False

name

Stacked Ensemble Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for stacked ensemble classes.

describe

Describe a component and its parameters.

feature_importance

Not implemented for StackedEnsembleClassifier and StackedEnsembleRegressor.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for stacked ensemble classes.

Returns

default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Not implemented for StackedEnsembleClassifier and StackedEnsembleRegressor.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.StandardScaler(random_seed=0, **kwargs)[source]

A transformer that standardizes input features by removing the mean and scaling to unit variance.

Parameters

random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{}

modifies_features

True

modifies_target

False

name

Standard Scaler

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

fit

Fits component to data.

fit_transform

Fit and transform data using the standard scaler component.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

save

Saves component at file path.

transform

Transform data using the fitted standard scaler.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

fit(self, X, y=None)

Fits component to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features]

  • y (pd.Series, optional) – The target training data of length [n_samples]

Returns

self

Raises

MethodPropertyNotFoundError – If component does not have a fit method or a component_obj that implements fit.

fit_transform(self, X, y=None)[source]

Fit and transform data using the standard scaler component.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

Transformed data.

Return type

pd.DataFrame

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

transform(self, X, y=None)[source]

Transform data using the fitted standard scaler.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

Transformed data.

Return type

pd.DataFrame

class evalml.pipelines.SVMClassifier(C=1.0, kernel='rbf', gamma='auto', probability=True, random_seed=0, **kwargs)[source]

Support Vector Machine Classifier.

Parameters
  • C (float) – The regularization parameter. The strength of the regularization is inversely proportional to C. Must be strictly positive. The penalty is a squared l2 penalty. Defaults to 1.0.

  • kernel ({"poly", "rbf", "sigmoid"}) – Specifies the kernel type to be used in the algorithm. Defaults to “rbf”.

  • gamma ({"scale", "auto"} or float) – Kernel coefficient for “rbf”, “poly” and “sigmoid”. Defaults to “auto”. - If gamma=’scale’ is passed then it uses 1 / (n_features * X.var()) as value of gamma - If “auto” (default), uses 1 / n_features

  • probability (boolean) – Whether to enable probability estimates. Defaults to True.

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “C”: Real(0, 10), “kernel”: [“poly”, “rbf”, “sigmoid”], “gamma”: [“scale”, “auto”],}

model_family

ModelFamily.SVM

modifies_features

True

modifies_target

False

name

SVM Classifier

supported_problem_types

[ ProblemTypes.BINARY, ProblemTypes.MULTICLASS, ProblemTypes.TIME_SERIES_BINARY, ProblemTypes.TIME_SERIES_MULTICLASS,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance only works with linear kernels.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance only works with linear kernels.

If the kernel isn’t linear, we return a numpy array of zeros.

Returns

Feature importance of fitted SVM classifier or a numpy array of zeroes if the kernel is not linear.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.SVMRegressor(C=1.0, kernel='rbf', gamma='auto', random_seed=0, **kwargs)[source]

Support Vector Machine Regressor.

Parameters
  • C (float) – The regularization parameter. The strength of the regularization is inversely proportional to C. Must be strictly positive. The penalty is a squared l2 penalty. Defaults to 1.0.

  • kernel ({"poly", "rbf", "sigmoid"}) – Specifies the kernel type to be used in the algorithm. Defaults to “rbf”.

  • gamma ({"scale", "auto"} or float) – Kernel coefficient for “rbf”, “poly” and “sigmoid”. Defaults to “auto”. - If gamma=’scale’ is passed then it uses 1 / (n_features * X.var()) as value of gamma - If “auto” (default), uses 1 / n_features

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{ “C”: Real(0, 10), “kernel”: [“poly”, “rbf”, “sigmoid”], “gamma”: [“scale”, “auto”],}

model_family

ModelFamily.SVM

modifies_features

True

modifies_target

False

name

SVM Regressor

supported_problem_types

[ ProblemTypes.REGRESSION, ProblemTypes.TIME_SERIES_REGRESSION,]

training_only

False

Methods

clone

Constructs a new component with the same parameters and random state.

default_parameters

Returns the default parameters for this component.

describe

Describe a component and its parameters.

feature_importance

Feature importance of fitted SVM regresor.

fit

Fits estimator to data.

load

Loads component at file path.

needs_fitting

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

parameters

Returns the parameters which were used to initialize the component.

predict

Make predictions using selected features.

predict_proba

Make probability estimates for labels.

save

Saves component at file path.

clone(self)

Constructs a new component with the same parameters and random state.

Returns

A new instance of this component with identical parameters and random state.

default_parameters(cls)

Returns the default parameters for this component.

Our convention is that Component.default_parameters == Component().parameters.

Returns

Default parameters for this component.

Return type

dict

describe(self, print_name=False, return_dict=False)

Describe a component and its parameters.

Parameters
  • print_name (bool, optional) – whether to print name of component

  • return_dict (bool, optional) – whether to return description as dictionary in the format {“name”: name, “parameters”: parameters}

Returns

Returns dictionary if return_dict is True, else None.

Return type

None or dict

property feature_importance(self)

Feature importance of fitted SVM regresor.

Only works with linear kernels. If the kernel isn’t linear, we return a numpy array of zeros.

Returns

The feature importance of the fitted SVM regressor, or an array of zeroes if the kernel is not linear.

fit(self, X, y=None)

Fits estimator to data.

Parameters
  • X (pd.DataFrame) – The input training data of shape [n_samples, n_features].

  • y (pd.Series, optional) – The target training data of length [n_samples].

Returns

self

static load(file_path)

Loads component at file path.

Parameters

file_path (str) – Location to load file.

Returns

ComponentBase object

needs_fitting(self)

Returns boolean determining if component needs fitting before calling predict, predict_proba, transform, or feature_importances.

This can be overridden to False for components that do not need to be fit or whose fit methods do nothing.

Returns

True.

property parameters(self)

Returns the parameters which were used to initialize the component.

predict(self, X)

Make predictions using selected features.

Parameters

X (pd.DataFrame) – Data of shape [n_samples, n_features].

Returns

Predicted values.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict method or a component_obj that implements predict.

predict_proba(self, X)

Make probability estimates for labels.

Parameters

X (pd.DataFrame) – Features.

Returns

Probability estimates.

Return type

pd.Series

Raises

MethodPropertyNotFoundError – If estimator does not have a predict_proba method or a component_obj that implements predict_proba.

save(self, file_path, pickle_protocol=cloudpickle.DEFAULT_PROTOCOL)

Saves component at file path.

Parameters
  • file_path (str) – Location to save file.

  • pickle_protocol (int) – The pickle data stream format.

class evalml.pipelines.TargetEncoder(cols=None, smoothing=1.0, handle_unknown='value', handle_missing='value', random_seed=0, **kwargs)[source]

A transformer that encodes categorical features into target encodings.

Parameters
  • cols (list) – Columns to encode. If None, all string columns will be encoded, otherwise only the columns provided will be encoded. Defaults to None

  • smoothing (float) – The smoothing factor to apply. The larger this value is, the more influence the expected target value has on the resulting target encodings. Must be strictly larger than 0. Defaults to 1.0

  • handle_unknown (string) – Determines how to handle unknown categories for a feature encountered. Options are ‘value’, ‘error’, nd ‘return_nan’. Defaults to ‘value’, which replaces with the target mean

  • handle_missing (string) – Determines how to handle missing values encountered during fit or transform. Options are ‘value’, ‘error’, and ‘return_nan’. Defaults to ‘value’, which replaces with the target mean

  • random_seed (int) – Seed for the random number generator. Defaults to 0.

Attributes

hyperparameter_ranges

{}

modifies_features

True

modifies_target

False

name

Target Encoder

training_only

False

Methods