time_series_featurizer#
Transformer that delays input features and target variable for time series problems.
Module Contents#
Classes Summary#
Transformer that delays input features and target variable for time series problems. |
Contents#
- class evalml.pipelines.components.transformers.preprocessing.time_series_featurizer.TimeSeriesFeaturizer(time_index=None, max_delay=2, gap=0, forecast_horizon=1, conf_level=0.05, rolling_window_size=0.25, delay_features=True, delay_target=True, random_seed=0, **kwargs)[source]#
Transformer that delays input features and target variable for time series problems.
This component uses an algorithm based on the autocorrelation values of the target variable to determine which lags to select from the set of all possible lags.
The algorithm is based on the idea that the local maxima of the autocorrelation function indicate the lags that have the most impact on the present time.
The algorithm computes the autocorrelation values and finds the local maxima, called “peaks”, that are significant at the given conf_level. Since lags in the range [0, 10] tend to be predictive but not local maxima, the union of the peaks is taken with the significant lags in the range [0, 10]. At the end, only selected lags in the range [0, max_delay] are used.
Parametrizing the algorithm by conf_level lets the AutoMLAlgorithm tune the set of lags chosen so that the chances of finding a good set of lags is higher.
Using conf_level value of 1 selects all possible lags.
- Parameters
time_index (str) – Name of the column containing the datetime information used to order the data. Ignored.
max_delay (int) – Maximum number of time units to delay each feature. Defaults to 2.
forecast_horizon (int) – The number of time periods the pipeline is expected to forecast.
conf_level (float) – Float in range (0, 1] that determines the confidence interval size used to select which lags to compute from the set of [1, max_delay]. A delay of 1 will always be computed. If 1, selects all possible lags in the set of [1, max_delay], inclusive.
rolling_window_size (float) – Float in range (0, 1] that determines the size of the window used for rolling features. Size is computed as rolling_window_size * max_delay.
delay_features (bool) – Whether to delay the input features. Defaults to True.
delay_target (bool) – Whether to delay the target. Defaults to True.
gap (int) – The number of time units between when the features are collected and when the target is collected. For example, if you are predicting the next time step’s target, gap=1. This is only needed because when gap=0, we need to be sure to start the lagging of the target variable at 1. Defaults to 1.
random_seed (int) – Seed for the random number generator. This transformer performs the same regardless of the random seed provided.
Attributes
hyperparameter_ranges
Real(0.001, 1.0), “rolling_window_size”: Real(0.001, 1.0)}:type: {“conf_level”
modifies_features
True
modifies_target
False
name
Time Series Featurizer
needs_fitting
True
target_colname_prefix
target_delay_{}
training_only
False
Methods
Constructs a new component with the same parameters and random state.
Returns the default parameters for this component.
Describe a component and its parameters.
Fits the DelayFeatureTransformer.
Fit the component and transform the input data.
Loads component at file path.
Returns the parameters which were used to initialize the component.
Saves component at file path.
Computes the delayed values and rolling means for X and y.
- 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 DelayFeatureTransformer.
- 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
- Raises
ValueError – if self.time_index is None
- fit_transform(self, X, y=None)[source]#
Fit the component and transform the input data.
- Parameters
X (pd.DataFrame) – Data to transform.
y (pd.Series, or None) – Target.
- 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
- 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 delayed values and rolling means for X and y.
The chosen delays are determined by the autocorrelation function of the target variable. See the class docstring for more information on how they are chosen. If y is None, all possible lags are chosen.
If y is not None, it will also compute the delayed values for the target variable.
The rolling means for all numeric features in X and y, if y is numeric, are also returned.
- Parameters
X (pd.DataFrame or None) – Data to transform. None is expected when only the target variable is being used.
y (pd.Series, or None) – Target.
- Returns
Transformed X. No original features are returned.
- Return type
pd.DataFrame