Start#
In this guide, we’ll show how you can use EvalML to automatically find the best pipeline for predicting whether or not a credit card transaction is fradulent. Along the way, we’ll highlight EvalML’s built-in tools and features for understanding and interacting with the search process.
[1]:
import evalml
from evalml import AutoMLSearch
from evalml.utils import infer_feature_types
pandas.core.index is deprecated and will be removed in a future version. The public classes are available in the top-level namespace.
First, we load in the features and outcomes we want to use to train our model.
[2]:
X, y = evalml.demos.load_fraud(n_rows=250)
Number of Features
Boolean 1
Categorical 6
Numeric 5
Number of training examples: 250
Targets
False 88.40%
True 11.60%
Name: fraud, dtype: object
First, we will clean the data. Since EvalML accepts a pandas input, it can run type inference on this data directly. Since we’d like to change the types inferred by EvalML, we can use the infer_feature_types utility method. Here’s what we’re going to do with the following dataset:
Reformat the
expiration_datecolumn so it reflects a more familiar date format.Cast the
latandlngcolumns from float to str.Use
infer_feature_typesto specify what types certain columns should be. For example, to avoid having theprovidercolumn be inferred as natural language text, we have specified it as a categorical column instead.
The infer_feature_types utility method takes a pandas or numpy input and converts it to a pandas dataframe with a Woodwork accessor, providing us with flexibility to cast the data as necessary.
[3]:
X.ww["expiration_date"] = X["expiration_date"].apply(
lambda x: "20{}-01-{}".format(x.split("/")[1], x.split("/")[0])
)
X = infer_feature_types(
X,
feature_types={
"store_id": "categorical",
"expiration_date": "datetime",
"lat": "categorical",
"lng": "categorical",
"provider": "categorical",
},
)
X.ww
[3]:
| Physical Type | Logical Type | Semantic Tag(s) | |
|---|---|---|---|
| Column | |||
| card_id | int64 | Integer | ['numeric'] |
| store_id | int64 | Integer | ['numeric'] |
| datetime | datetime64[ns] | Datetime | [] |
| amount | int64 | Integer | ['numeric'] |
| currency | string | Unknown | [] |
| customer_present | bool | Boolean | [] |
| expiration_date | datetime64[ns] | Datetime | [] |
| provider | category | Categorical | ['category'] |
| lat | float64 | Double | ['numeric'] |
| lng | float64 | Double | ['numeric'] |
| region | category | Categorical | ['category'] |
| country | category | Categorical | ['category'] |
In order to validate the results of the pipeline creation and optimization process, we will save some of our data as a holdout set.
[4]:
X_train, X_holdout, y_train, y_holdout = evalml.preprocessing.split_data(
X, y, problem_type="binary", test_size=0.2
)
Note: To provide data to EvalML, it is recommended that you initialize a woodwork accessor so that you control how EvalML will treat each feature, such as as a numeric feature, a categorical feature, a text feature or other type of feature. Consult the the Woodwork project for help on how to do this. Here, split_data() returns dataframes with woodwork accessors.
EvalML has many options to configure the pipeline search. At the minimum, we need to define an objective function. For simplicity, we will use the F1 score in this example. However, the real power of EvalML is in using domain-specific objective functions or building your own.
Below EvalML utilizes Bayesian optimization (EvalML’s default optimizer) to search and find the best pipeline defined by the given objective.
EvalML provides a number of parameters to control the search process. max_batches is one of the parameters which controls the stopping criterion for the AutoML search. It indicates the maximum number of rounds of AutoML to evaluate, where each round may train and score a variable number of pipelines. In this example, max_batches is set to 1.
** Graphing methods, like AutoMLSearch, on Jupyter Notebook and Jupyter Lab require ipywidgets to be installed.
** If graphing on Jupyter Lab, jupyterlab-plotly required. To download this, make sure you have npm installed.
[5]:
automl = AutoMLSearch(
X_train=X_train,
y_train=y_train,
problem_type="binary",
objective="f1",
max_batches=3,
verbose=True,
)
AutoMLSearch will use mean CV score to rank pipelines.
Removing columns ['currency'] because they are of 'Unknown' type
When we call search(), the search for the best pipeline will begin. There is no need to wrangle with missing data or categorical variables as EvalML includes various preprocessing steps (like imputation, one-hot encoding, feature selection) to ensure you’re getting the best results. As long as your data is in a single table, EvalML can handle it. If not, you can reduce your data to a single table by utilizing Featuretools and its Entity Sets.
You can find more information on pipeline components and how to integrate your own custom pipelines into EvalML here.
[6]:
automl.search(interactive_plot=False)
*****************************
* Beginning pipeline search *
*****************************
Optimizing for F1.
Greater score is better.
Using SequentialEngine to train and score pipelines.
Searching up to 3 batches for a total of None pipelines.
Allowed model families:
Evaluating Baseline Pipeline: Mode Baseline Binary Classification Pipeline
Mode Baseline Binary Classification Pipeline:
Starting cross validation
Finished cross validation - mean F1: 0.000
*****************************
* Evaluating Batch Number 1 *
*****************************
Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler:
Starting cross validation
Finished cross validation - mean F1: 0.399
Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.588
*****************************
* Evaluating Batch Number 2 *
*****************************
Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler + RF Classifier Select From Model:
Starting cross validation
Finished cross validation - mean F1: 0.280
Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + RF Classifier Select From Model:
Starting cross validation
Finished cross validation - mean F1: 0.677
*****************************
* Evaluating Batch Number 3 *
*****************************
Decision Tree Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.358
LightGBM Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.520
Extra Trees Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.310
Elastic Net Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Standard Scaler + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Standard Scaler + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.395
CatBoost Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.398
XGBoost Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler:
Starting cross validation
Finished cross validation - mean F1: 0.643
Search finished after 00:29
Best pipeline: Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + RF Classifier Select From Model
Best pipeline F1: 0.676923
[6]:
{1: {'Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler': '00:04',
'Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler': '00:02',
'Total time of batch': '00:07'},
2: {'Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler + RF Classifier Select From Model': '00:03',
'Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + RF Classifier Select From Model': '00:02',
'Total time of batch': '00:06'},
3: {'Decision Tree Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:02',
'LightGBM Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:02',
'Extra Trees Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:03',
'Elastic Net Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Standard Scaler + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Standard Scaler + Oversampler': '00:02',
'CatBoost Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + Oversampler': '00:01',
'XGBoost Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:02',
'Total time of batch': '00:16'}}
If you would like to suppress stdout output, set verbose=False. This is also the default behavior for AutoMLSearch if verbose is not specified.
Also, if you would like to see the interactive plot update dynamically over time as the search progresses, either remove the parameter or set interactive_plot=True. This is the default setting for search() if interactive_plot is not specified (it is set to False here due to documentation workaround).
[7]:
automl = AutoMLSearch(
X_train=X_train,
y_train=y_train,
problem_type="binary",
objective="f1",
max_batches=3,
verbose=False,
)
automl.search()
[7]:
{1: {'Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler': '00:02',
'Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler': '00:02',
'Total time of batch': '00:05'},
2: {'Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler + RF Classifier Select From Model': '00:02',
'Random Forest Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + RF Classifier Select From Model': '00:02',
'Total time of batch': '00:06'},
3: {'Decision Tree Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:02',
'LightGBM Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:02',
'Extra Trees Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:03',
'Elastic Net Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Standard Scaler + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Standard Scaler + Oversampler': '00:02',
'CatBoost Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + Oversampler': '00:01',
'XGBoost Classifier w/ Label Encoder + Select Columns By Type Transformer + Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + Select Columns Transformer + Select Columns Transformer + Label Encoder + Replace Nullable Types Transformer + Imputer + One Hot Encoder + Oversampler': '00:02',
'Total time of batch': '00:15'}}
We also provide a standalone search method which does all of the above in a single line, and returns the AutoMLSearch instance and data check results. If there were data check errors, AutoML will not be run and no AutoMLSearch instance will be returned.
After the search is finished we can view all of the pipelines searched, ranked by score. Internally, EvalML performs cross validation to score the pipelines. If it notices a high variance across cross validation folds, it will warn you. EvalML also provides additional data checks to analyze your data to assist you in producing the best performing pipeline.
[8]:
automl.rankings
[8]:
| id | pipeline_name | search_order | validation_score | mean_cv_score | standard_deviation_cv_score | percent_better_than_baseline | high_variance_cv | parameters | |
|---|---|---|---|---|---|---|---|---|---|
| 0 | 4 | Random Forest Classifier w/ Label Encoder + Re... | 4 | 0.676923 | 0.676923 | 0.187162 | 67.692308 | False | {'Label Encoder': {'positive_label': None}, 'D... |
| 1 | 10 | XGBoost Classifier w/ Label Encoder + Select C... | 10 | 0.642735 | 0.642735 | 0.170495 | 64.273504 | False | {'Label Encoder': {'positive_label': None}, 'N... |
| 2 | 2 | Random Forest Classifier w/ Label Encoder + Re... | 2 | 0.587709 | 0.587709 | 0.233493 | 58.770859 | False | {'Label Encoder': {'positive_label': None}, 'D... |
| 3 | 6 | LightGBM Classifier w/ Label Encoder + Select ... | 6 | 0.519943 | 0.519943 | 0.087198 | 51.994302 | False | {'Label Encoder': {'positive_label': None}, 'N... |
| 4 | 1 | Logistic Regression Classifier w/ Label Encode... | 1 | 0.398944 | 0.398944 | 0.143024 | 39.894419 | False | {'Label Encoder': {'positive_label': None}, 'D... |
| 5 | 9 | CatBoost Classifier w/ Label Encoder + Select ... | 9 | 0.398485 | 0.398485 | 0.147733 | 39.848485 | False | {'Label Encoder': {'positive_label': None}, 'N... |
| 6 | 8 | Elastic Net Classifier w/ Label Encoder + Sele... | 8 | 0.395153 | 0.395153 | 0.183837 | 39.515251 | False | {'Label Encoder': {'positive_label': None}, 'N... |
| 7 | 5 | Decision Tree Classifier w/ Label Encoder + Se... | 5 | 0.358333 | 0.358333 | 0.232289 | 35.833333 | False | {'Label Encoder': {'positive_label': None}, 'N... |
| 8 | 7 | Extra Trees Classifier w/ Label Encoder + Sele... | 7 | 0.309690 | 0.309690 | 0.141393 | 30.969031 | False | {'Label Encoder': {'positive_label': None}, 'N... |
| 9 | 3 | Logistic Regression Classifier w/ Label Encode... | 3 | 0.280375 | 0.280375 | 0.082662 | 28.037489 | False | {'Label Encoder': {'positive_label': None}, 'D... |
| 10 | 0 | Mode Baseline Binary Classification Pipeline | 0 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | False | {'Label Encoder': {'positive_label': None}, 'B... |
If we are interested in see more details about the pipeline, we can view a summary description using the id from the rankings table:
[9]:
automl.describe_pipeline(3)
***************************************************************************************************************************************************************************************************************************************
* Logistic Regression Classifier w/ Label Encoder + Replace Nullable Types Transformer + Drop Columns Transformer + DateTime Featurizer + Imputer + One Hot Encoder + Oversampler + Standard Scaler + RF Classifier Select From Model *
***************************************************************************************************************************************************************************************************************************************
Problem Type: binary
Model Family: Linear
Pipeline Steps
==============
1. Label Encoder
* positive_label : None
2. Replace Nullable Types Transformer
3. Drop Columns Transformer
* columns : ['currency']
4. DateTime Featurizer
* features_to_extract : ['year', 'month', 'day_of_week', 'hour']
* encode_as_categories : False
* time_index : None
5. Imputer
* categorical_impute_strategy : most_frequent
* numeric_impute_strategy : mean
* boolean_impute_strategy : most_frequent
* categorical_fill_value : None
* numeric_fill_value : None
* boolean_fill_value : None
6. One Hot Encoder
* top_n : 10
* features_to_encode : None
* categories : None
* drop : if_binary
* handle_unknown : ignore
* handle_missing : error
7. Oversampler
* sampling_ratio : 0.25
* k_neighbors_default : 5
* n_jobs : -1
* sampling_ratio_dict : None
* categorical_features : [3, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43]
* k_neighbors : 5
8. Standard Scaler
9. RF Classifier Select From Model
* number_features : None
* n_estimators : 10
* max_depth : None
* percent_features : 0.5
* threshold : median
* n_jobs : -1
10. Logistic Regression Classifier
* penalty : l2
* C : 1.0
* n_jobs : -1
* multi_class : auto
* solver : lbfgs
Training
========
Training for binary problems.
Objective to optimize binary classification pipeline thresholds for: <evalml.objectives.standard_metrics.F1 object at 0x7f80885cb0a0>
Total training time (including CV): 2.9 seconds
Cross Validation
----------------
F1 MCC Binary Log Loss Binary Gini AUC Precision Balanced Accuracy Binary Accuracy Binary # Training # Validation
0 0.244 0.098 0.439 0.284 0.642 0.152 0.575 0.537 133 67
1 0.375 0.290 0.456 0.322 0.661 0.375 0.645 0.851 133 67
2 0.222 0.110 0.544 0.085 0.542 0.182 0.567 0.788 134 66
mean 0.280 0.166 0.480 0.230 0.615 0.236 0.596 0.725 - -
std 0.083 0.108 0.056 0.127 0.064 0.121 0.043 0.166 - -
coef of var 0.295 0.650 0.117 0.553 0.104 0.513 0.072 0.229 - -
We can also view the pipeline parameters directly:
[10]:
pipeline = automl.get_pipeline(3)
print(pipeline.parameters)
{'Label Encoder': {'positive_label': None}, 'Drop Columns Transformer': {'columns': ['currency']}, 'DateTime Featurizer': {'features_to_extract': ['year', 'month', 'day_of_week', 'hour'], 'encode_as_categories': False, 'time_index': None}, 'Imputer': {'categorical_impute_strategy': 'most_frequent', 'numeric_impute_strategy': 'mean', 'boolean_impute_strategy': 'most_frequent', 'categorical_fill_value': None, 'numeric_fill_value': None, 'boolean_fill_value': None}, 'One Hot Encoder': {'top_n': 10, 'features_to_encode': None, 'categories': None, 'drop': 'if_binary', 'handle_unknown': 'ignore', 'handle_missing': 'error'}, 'Oversampler': {'sampling_ratio': 0.25, 'k_neighbors_default': 5, 'n_jobs': -1, 'sampling_ratio_dict': None, 'categorical_features': [3, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43], 'k_neighbors': 5}, 'RF Classifier Select From Model': {'number_features': None, 'n_estimators': 10, 'max_depth': None, 'percent_features': 0.5, 'threshold': 'median', 'n_jobs': -1}, 'Logistic Regression Classifier': {'penalty': 'l2', 'C': 1.0, 'n_jobs': -1, 'multi_class': 'auto', 'solver': 'lbfgs'}}
We can now select the best pipeline and score it on our holdout data:
[11]:
pipeline = automl.best_pipeline
pipeline.score(X_holdout, y_holdout, ["f1"])
[11]:
OrderedDict([('F1', 0.6666666666666666)])
We can also visualize the structure of the components contained by the pipeline:
[12]:
pipeline.graph()
[12]:
