_OneSegmentLinearTrendBaseTransform¶

class _OneSegmentLinearTrendBaseTransform(in_column: str, regressor: sklearn.base.RegressorMixin, poly_degree: int = 1)[source]¶

Transform for one segment that implements trend subtraction and reconstruction feature.

Create instance of _OneSegmentLinearTrendBaseTransform.

Parameters

in_column (str) – name of processed column
regressor (sklearn.base.RegressorMixin) – instance of sklearn :py:class`sklearn.base.RegressorMixin` to predict trend
poly_degree (int) – degree of polynomial to fit trend on

Inherited-members

Methods

`fit`(df)	Fit regression detrend_model with data from df.
`fit_transform`(df)	Fit regression detrend_model with data from df and subtract the trend from df.
`inverse_transform`(df)	Inverse transformation for trend subtraction: add trend to prediction.
`set_params`(**params)	Return new object instance with modified parameters.
`to_dict`()	Collect all information about etna object in dict.
`transform`(df)	Transform data from df: subtract linear trend found by regressor.

Fit regression detrend_model with data from df.

Parameters: df (pandas.core.frame.DataFrame) – data that regressor should be trained with
Returns: instance with trained regressor
Return type: _OneSegmentLinearTrendBaseTransform

fit_transform(df: pandas.core.frame.DataFrame) → pandas.core.frame.DataFrame[source]¶

Fit regression detrend_model with data from df and subtract the trend from df.

Parameters: df (pandas.core.frame.DataFrame) – data to train regressor and transform
Returns: residue after trend subtraction
Return type: pd.DataFrame

inverse_transform(df: pandas.core.frame.DataFrame) → pandas.core.frame.DataFrame[source]¶

Inverse transformation for trend subtraction: add trend to prediction.

transform(df: pandas.core.frame.DataFrame) → pandas.core.frame.DataFrame[source]¶

Transform data from df: subtract linear trend found by regressor.

Parameters: df (pandas.core.frame.DataFrame) – data to subtract trend from
Returns: residue after trend subtraction
Return type: pd.DataFrame