STLTransform

class STLTransform(in_column: str, period: int, model: Union[str, statsmodels.tsa.base.tsa_model.TimeSeriesModel] = 'arima', robust: bool = False, model_kwargs: Optional[Dict[str, Any]] = None, stl_kwargs: Optional[Dict[str, Any]] = None)[source]

Bases: etna.transforms.base.PerSegmentWrapper

Transform that uses statsmodels.tsa.seasonal.STL to subtract season and trend from the data.

Warning

This transform can suffer from look-ahead bias. For transforming data at some timestamp it uses information from the whole train part.

Init STLTransform.

Parameters

  • in_column (str) – name of processed column

  • period (int) – size of seasonality

  • model (Union[str, statsmodels.tsa.base.tsa_model.TimeSeriesModel]) –

    model to predict trend, default options are:

    1. ”arima”: ARIMA(data, 1, 1, 0) (default)

    2. ”holt”: ETSModel(data, trend='add')

    Custom model should be a subclass of statsmodels.tsa.base.tsa_model.TimeSeriesModel and have method get_prediction (not just predict)

  • robust (bool) – flag indicating whether to use robust version of STL

  • model_kwargs (Optional[Dict[str, Any]]) – parameters for the model like in statsmodels.tsa.seasonal.STLForecast

  • stl_kwargs (Optional[Dict[str, Any]]) – additional parameters for statsmodels.tsa.seasonal.STLForecast

Inherited-members

Methods

fit(df)

Fit transform on each segment.

fit_transform(df)

May be reimplemented.

inverse_transform(df)

Apply inverse_transform to each segment.

transform(df)

Apply transform to each segment separately.