tsfeast package¶

Submodules¶

tsfeast.funcs module¶

Time series feature generator functions.

tsfeast.funcs.get_busdays_in_month(dt: pandas._libs.tslibs.timestamps.Timestamp) → int[source]¶

Get the number of business days in a month period, using US holidays.

Parameters: dt (pd.Timestamp) – Desired month.
Returns: int – Number of business days in the month.

tsfeast.funcs.get_datetime_features(data: Union[pandas.core.frame.DataFrame, pandas.core.series.Series], date_col: Optional[str] = None, dt_format: Optional[str] = None, freq: Optional[str] = None) → pandas.core.frame.DataFrame[source]¶

Get features based on datetime index, including year, month, week, weekday, quarter, days in month, business days in month and leap year.

Parameters

data (pd.DataFrame, pd.Series) – Original data.
date_col (Optional[str]) – Column name containing date/timestamp.
dt_format (Optional[str]) – Date/timestamp format, e.g. %Y-%m-%d for 2020-01-31.
freq (Optional[str]) – Date frequency.

Returns

pd.DataFrame – Date features.

tsfeast.funcs.get_lag_features(data: pandas.core.frame.DataFrame, n_lags: int) → pandas.core.frame.DataFrame[source]¶

Get n-lagged features for data.

Parameters

data (pd.DataFrame) – Original data.
n_lags (int) – Number of lags to generate.

Returns

pd.DataFrame – Lagged values of specified dataset.

tsfeast.funcs.get_rolling_features(data: pandas.core.frame.DataFrame, window_lengths: List[int]) → pandas.core.frame.DataFrame[source]¶

Get rolling metrics (mean, std, min, max) for each specified window length.

Parameters

data (pd.DataFrame) – Original data.
window_lengths (List[int]) – List of window lengths to generate.

Returns

pd.DataFrame – Rolling mean, std, min and max for each specified window length.

tsfeast.funcs.get_ewma_features(data: pandas.core.frame.DataFrame, window_lengths: List[int]) → pandas.core.frame.DataFrame[source]¶

Get an exponentially-weighted moving average for each specified window length.

Parameters

data (pd.DataFrame) – Original data.
window_lengths (List[int]) – List of window lengths to generate.

Returns

pd.DataFrame – Exponentially-weighted moving average for each specified window length.

tsfeast.funcs.get_change_features(data: pandas.core.frame.DataFrame, period_lengths: List[int]) → pandas.core.frame.DataFrame[source]¶

Get percent change for all features for each specified period length.

Parameters

data (pd.DataFrame) – Original data.
period_lengths (List[int]) – A list of period lengths to generate.

Returns

pd.DataFrame – Percent changes for all features.

tsfeast.funcs.get_difference_features(data: pandas.core.frame.DataFrame, n_diffs: int) → pandas.core.frame.DataFrame[source]¶

Get n differences for all features.

Parameters

data (pd.DataFrame) – Original data.
n_diffs (int) – Number of differences to return.

Returns

pd.DataFrame – N-differenced data.

tsfeast.metrics module¶

Custom scoring metrics.

tsfeast.metrics.bic_score(mse: float, n: int, p: int)[source]¶

Calcuate BIC score.

Parameters

mse (float) – Mean-squared error.
n (int) – Number of observations.
p (int) – Number of parameters

Returns

float – BIC value.

tsfeast.metrics.bic_scorer(estimator: tsfeast._base.BaseContainer, X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray])[source]¶: Score SciKit-Learn estimator using BIC.

tsfeast.models module¶

Module for Scikit-Learn Regressor with ARMA Residuals and Scikit-Learn API wrapper for Statsmodels TSA models.

class tsfeast.models.ARMARegressor(estimator: sklearn.linear_model._base.LinearModel = LinearRegression(), order: Tuple[int, int, int] = (1, 0, 0))[source]¶

Bases: tsfeast._base.BaseContainer

Estimator for Scikit-Learn estimator with ARMA residuals.

Parameters

estimator (LinearModel) – Scikit-Learn linear estimator.
order (Tuple[int, int, int]) – ARIMA order for residuals.

estimator¶

The Scikit-Learn regressor.

Type: LinearModel

order¶

The (p,d,q,) order of the ARMA model.

Type: Tuple[int, int, int]

intercept_¶

The fitted estimator’s intercept.

Type: float

coef_¶

The fitted estimator’s coefficients.

Type: np.ndarray

arma_¶

The fitted ARMA model.

Type: ARIMA

fitted_values_¶

The combined estimator and ARMA fitted values.

Type: np.ndarray

resid_¶

The combined estimator and ARMA residual values.

Type: np.ndarray

Instantiate ARMARegressor object.

Parameters

estimator (LinearRegression) – Scikit-Learn linear estimator.
order (Tuple[int, int, int]) – ARIMA order for residuals.

__init__(estimator: sklearn.linear_model._base.LinearModel = LinearRegression(), order: Tuple[int, int, int] = (1, 0, 0))[source]¶

Instantiate ARMARegressor object.

Parameters

estimator (LinearRegression) – Scikit-Learn linear estimator.
order (Tuple[int, int, int]) – ARIMA order for residuals.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]) → tsfeast._base.BaseContainer¶

Fit the estimator.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

Returns

BaseContainer – Self.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

predict(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Make predictions with fitted estimator.

Parameters: X (array of shape [n_samples, n_features]) – The input samples.
Returns: np.ndarray – Array of predicted values.

score(X, y, sample_weight=None)¶

Return the coefficient of determination R^2 of the prediction.

The coefficient R^2 is defined as (1 - \frac{u}{v}), where u is the residual sum of squares ((y_true - y_pred) ** 2).sum() and v is the total sum of squares ((y_true - y_true.mean()) ** 2).sum(). The best possible score is 1.0 and it can be negative (because the model can be arbitrarily worse). A constant model that always predicts the expected value of y, disregarding the input features, would get a R^2 score of 0.0.

Parameters

X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.
y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.
sample_weight (array-like of shape (n_samples,), default=None) – Sample weights.

Returns

score (float) – R^2 of self.predict(X) wrt. y.

Notes

The R^2 score used when calling score on a regressor uses multioutput='uniform_average' from version 0.23 to keep consistent with default value of r2_score(). This influences the score method of all the multioutput regressors (except for MultiOutputRegressor).

set_params(**params)¶

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

class tsfeast.models.TSARegressor(model: statsmodels.base.model.Model, use_exog: bool = False, **kwargs)[source]¶

Bases: tsfeast._base.BaseContainer

Estimator for StatsModels TSA model.

Parameters

model (Model) – An uninstantiated Statsmodels TSA model.
use_exog (bool) – Whether to use exogenous features; default False.
kwargs – Additional kwargs for Statsmodels model.

fitted_model_¶

The fitted Statmodels model object.

Type: Model

summary_¶: The fitted Statmodels model summary results.

Instantiate TSARegressor object.

model: Model: An uninstantiated Statsmodels TSA model.
use_exog: bool: Whether to use exogenous features; default False.
kwargs:: Additional kwargs for Statsmodels model.

__init__(model: statsmodels.base.model.Model, use_exog: bool = False, **kwargs)[source]¶

Instantiate TSARegressor object.

model: Model: An uninstantiated Statsmodels TSA model.
use_exog: bool: Whether to use exogenous features; default False.
kwargs:: Additional kwargs for Statsmodels model.

Fit the estimator.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

Returns

BaseContainer – Self.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

predict(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Make predictions with fitted estimator.

Parameters: X (array of shape [n_samples, n_features]) – The input samples.
Returns: np.ndarray – Array of predicted values.

score(X, y, sample_weight=None)¶

Return the coefficient of determination R^2 of the prediction.

Parameters

X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.
y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.
sample_weight (array-like of shape (n_samples,), default=None) – Sample weights.

Returns

score (float) – R^2 of self.predict(X) wrt. y.

Notes

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

tsfeast.splitter module¶

Time Series Windows Module.

Note These classes split data into n, equal-length, sliding training/test windows. This differs from Scikit-Learn’s TimeSeriesSplit implementation where windows are accumulated:

Scikit-Learn¶

Win 0 |----| Win 1 |--------| Win2 |------------|

TimeSeriesWindows¶

Win 0 |----|——– Win 1 -|----|——- Win2 –|----|——

class tsfeast.splitter.TimeSeriesWindows(train_length: int, test_length: int, gap_length: int = 0)[source]¶

Bases: object

split(y: pandas.core.frame.DataFrame, x: pandas.core.frame.DataFrame) → List[pandas.core.frame.DataFrame][source]¶

class tsfeast.splitter.EndogSeriesWindows(min_train_length: int, test_length: int, max_train_length: Optional[int] = None, gap_length: int = 0)[source]¶

Bases: tsfeast.splitter.TimeSeriesWindows

split(y: pandas.core.frame.DataFrame, x=None) → List[pandas.core.frame.DataFrame][source]¶

tsfeast.transformers module¶

Time series feature generators as Scikit-Learn compatible transformers.

class tsfeast.transformers.BaseTransformer(fillna: bool = True)[source]¶

Bases: sklearn.base.BaseEstimator, sklearn.base.TransformerMixin

Base transformer object.

Instantiate transformer object.

__init__(fillna: bool = True)[source]¶: Instantiate transformer object.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray][source]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

Scikit-Learn Pipelines only call the .transform() method during the .predict() method, which is appropriate to prevent data leakage in predictions. However, most of the transformers in this module take a set of features and generate new features; there’s no inherent method to transform some timeseries features given a fitted estimator.

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

get_feature_names() → List[str][source]¶: Get list of feature names.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer[source]¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

class tsfeast.transformers.OriginalFeatures(fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Return original features.

Instantiate transformer object.

__init__(fillna: bool = True)¶: Instantiate transformer object.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.Scaler[source]¶

Bases: tsfeast.transformers.BaseTransformer

Wrap StandardScaler to maintain column names.

Instantiate transformer object.

__init__()[source]¶: Instantiate transformer object.

fit(X: pandas.core.frame.DataFrame, y=None) → tsfeast.transformers.Scaler[source]¶

Fit transformer object to data.

Parameters

X (pd.DataFrame) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Transformed features.

transform(X: pandas.core.frame.DataFrame, y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray][source]¶

Fit transformer object to data.

Parameters

X (pd.DataFrame) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Transformed features.

inverse_transform(X: pandas.core.frame.DataFrame, copy: bool = True) → pandas.core.frame.DataFrame[source]¶

Transform scaled data into original feature space.

Parameters

X (pd.DataFrame) – The input samples.
copy (bool) – Default True; if False, try to avoid a copy and do inplace scaling instead.

Returns

Data – Data in original feature space.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

class tsfeast.transformers.DateTimeFeatures(date_col: Optional[str] = None, dt_format: Optional[str] = None, freq: Optional[str] = None)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate datetime features.

Instantiate transformer object.

date_col: Optional[str]: Column name containing date/timestamp.
dt_format: Optional[str]: Date/timestamp format, e.g. %Y-%m-%d for 2020-01-31.

__init__(date_col: Optional[str] = None, dt_format: Optional[str] = None, freq: Optional[str] = None)[source]¶

Instantiate transformer object.

date_col: Optional[str]: Column name containing date/timestamp.
dt_format: Optional[str]: Date/timestamp format, e.g. %Y-%m-%d for 2020-01-31.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.DateTimeFeatures[source]¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.LagFeatures(n_lags: int, fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate lag features.

Instantiate transformer object.

Parameters: n_lags (int) – Number of lags to generate.

__init__(n_lags: int, fillna: bool = True)[source]¶

Instantiate transformer object.

Parameters: n_lags (int) – Number of lags to generate.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.RollingFeatures(window_lengths: List[int], fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate rolling features.

Instantiate transformer object.

Parameters: window_lengths (L:ist[int]) – Length of window(s) to create.

__init__(window_lengths: List[int], fillna: bool = True)[source]¶

Instantiate transformer object.

Parameters: window_lengths (L:ist[int]) – Length of window(s) to create.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.EwmaFeatures(window_lengths: List[int], fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate exponentially-weighted moving-average features.

Instantiate transformer object.

Parameters: window_lengths (L:ist[int]) – Length of window(s) to create.

__init__(window_lengths: List[int], fillna: bool = True)[source]¶

Instantiate transformer object.

Parameters: window_lengths (L:ist[int]) – Length of window(s) to create.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.ChangeFeatures(period_lengths: List[int], fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate period change features.

Instantiate transformer object.

Parameters: period_lengths (List[int]) – Length of period[s] to generate change features.

__init__(period_lengths: List[int], fillna: bool = True)[source]¶

Instantiate transformer object.

Parameters: period_lengths (List[int]) – Length of period[s] to generate change features.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.DifferenceFeatures(n_diffs: int, fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate difference features.

Instantiate transformer object.

Parameters: n_diffs (int) – Number of differences to calculate.

__init__(n_diffs: int, fillna: bool = True)[source]¶

Instantiate transformer object.

Parameters: n_diffs (int) – Number of differences to calculate.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.PolyFeatures(degree=2)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate polynomial features.

Instantiate transformer object.

Parameters: degree (int) – Degree of polynomial to use.

__init__(degree=2)[source]¶

Instantiate transformer object.

Parameters: degree (int) – Degree of polynomial to use.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

class tsfeast.transformers.InteractionFeatures(fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Wrap PolynomialFeatures to extract interactions and keep column names.

Instantiate transformer object.

__init__(fillna: bool = True)¶: Instantiate transformer object.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

tsfeast.tsfeatures module¶

Time series features module.

class tsfeast.tsfeatures.TimeSeriesFeatures(datetime: str, trend: str = 'n', lags: Optional[int] = None, rolling: Optional[List[int]] = None, ewma: Optional[List[int]] = None, pct_chg: Optional[List[int]] = None, diffs: Optional[int] = None, polynomial: Optional[int] = None, interactions: bool = True, fillna: bool = True)[source]¶

Bases: tsfeast.transformers.BaseTransformer

Generate multiple time series feature in one transformer.

Instanatiate transformer object.

__init__(datetime: str, trend: str = 'n', lags: Optional[int] = None, rolling: Optional[List[int]] = None, ewma: Optional[List[int]] = None, pct_chg: Optional[List[int]] = None, diffs: Optional[int] = None, polynomial: Optional[int] = None, interactions: bool = True, fillna: bool = True)[source]¶: Instanatiate transformer object.

fit(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → tsfeast.transformers.BaseTransformer¶

Fit transformer object to data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

BaseTransformer – Self.

fit_transform(X, y=None, **fit_params)¶

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters

X (array-like of shape (n_samples, n_features)) – Input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
**fit_params (dict) – Additional fit parameters.

Returns

X_new (ndarray array of shape (n_samples, n_features_new)) – Transformed array.

get_feature_names() → List[str]¶: Get list of feature names.

get_params(deep=True)¶

Get parameters for this estimator.

Parameters: deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
Returns: params (dict) – Parameter names mapped to their values.

set_params(**params)¶

Set the parameters of this estimator.

Parameters: **params (dict) – Estimator parameters.
Returns: self (estimator instance) – Estimator instance.

transform(X: Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray], y=None) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]¶

Transform fitted data.

Parameters

X (array of shape [n_samples, n_features]) – The input samples.
y (None) – Not used; included for compatibility, only.

Returns

Data – Array-like object of transformed data.

Notes

For time series lags, changes, etc., we have access to past data for feature generation without risk of data leakage; certain features (e.g. lags) require this to avoid NaNs or zeros.

We append new X to our original features and transform on entire dataset, keeping only the last n rows. Appropriate for time series transformations, only.

tsfeast.utils module¶

Miscellaneous utility functions.

tsfeast.utils.to_list(x: Union[int, List]) → List[int][source]¶: Ensure parameter is list of integer(s).

tsfeast.utils.array_to_dataframe(x: numpy.ndarray) → pandas.core.frame.DataFrame[source]¶: Convert Numpy array to Pandas DataFrame with default column names.

tsfeast.utils.array_to_series(x: numpy.ndarray) → pandas.core.series.Series[source]¶: Convert Numpy array to Pandas Series with default name.

tsfeast.utils.plot_diag(residuals: Optional[Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]] = None, estimator: Optional[sklearn.linear_model._base.LinearModel] = None, X: Optional[Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]] = None, y: Optional[Union[pandas.core.frame.DataFrame, pandas.core.series.Series, numpy.ndarray]] = None)[source]¶

Plot regression diagnostics.

Generate residuals plot, QQ plot, ACF plot and PACF plot, given either an array-like object of residuals or and estimator and X and y data arrays.

Parameters

residuals (Data) – Model residual errors.
estimator (LinearModel) – Scikit-Learn generalized linear model.
X (array of shape [n_samples, n_features]) – The input samples.
y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).

Raises

ValueError –

If neither residuals or estimator provider. - If estimator provided without X and y data.

tsfeast package¶

Submodules¶

tsfeast.funcs module¶

tsfeast.metrics module¶

tsfeast.models module¶

tsfeast.splitter module¶

Scikit-Learn¶

TimeSeriesWindows¶

tsfeast.transformers module¶

tsfeast.tsfeatures module¶

tsfeast.utils module¶

Module contents¶