GroupNormalizer#

class pytorch_forecasting.data.encoders.GroupNormalizer(method: str = 'standard', groups: List[str] = [], center: bool = True, scale_by_group: bool = False, transformation: str | Tuple[Callable, Callable] | None = None, method_kwargs: Dict[str, Any] = {})[source]#

Bases: TorchNormalizer

Normalizer that scales by groups.

For each group a scaler is fitted and applied. This scaler can be used as target normalizer or also to normalize any other variable.

Group normalizer to normalize a given entry by groups. Can be used as target normalizer.

Parameters:

  • method (str, optional) – method to rescale series. Either “standard” (standard scaling) or “robust” (scale using quantiles 0.25-0.75). Defaults to “standard”.

  • method_kwargs (Dict[str, Any], optional) – Dictionary of method specific arguments as listed below * “robust” method: “upper”, “lower”, “center” quantiles defaulting to 0.75, 0.25 and 0.5

  • groups (List[str], optional) – Group names to normalize by. Defaults to [].

  • center (bool, optional) – If to center the output to zero. Defaults to True.

  • scale_by_group (bool, optional) – If to scale the output by group, i.e. norm is calculated as (group1_norm * group2_norm * ...) ^ (1 / n_groups). Defaults to False.

  • transformation (Union[str, Tuple[Callable, Callable]] optional) –

    Transform values before applying normalizer. Available options are

    • None (default): No transformation of values

    • log: Estimate in log-space leading to a multiplicative model

    • log1p: Estimate in log-space but add 1 to values before transforming for stability

      (e.g. if many small values <<1 are present). Note, that inverse transform is still only torch.exp() and not torch.expm1().

    • logit: Apply logit transformation on values that are between 0 and 1

    • count: Apply softplus to output (inverse transformation) and x + 1 to input

      (transformation)

    • softplus: Apply softplus to output (inverse transformation) and inverse softplus to input

      (transformation)

    • relu: Apply max(0, x) to output

    • Dict[str, Callable] of PyTorch functions that transforms and inversely transforms values. forward and reverse entries are required. inverse transformation is optional and should be defined if reverse is not the inverse of the forward transformation. inverse_torch can be defined to provide a torch distribution transform for inverse transformations.

Inherited-members:

Methods

extra_repr()

fit(y, X)

Determine scales for each group

fit_transform(y, X[, return_norm])

Fit normalizer and scale input data.

get_metadata_routing()

Get metadata routing of this object.

get_norm(X)

Get scaling parameters for multiple groups.

get_parameters(groups[, group_names])

Get fitted scaling parameters for a given group.

get_params([deep])

Get parameters for this estimator.

get_transform(transformation)

Return transformation functions.

inverse_preprocess(y)

Inverse preprocess re-scaled data (e.g.

inverse_transform(y, X)

Rescaling data to original scale - not implemented - call class with target scale instead.

preprocess(y)

Preprocess input data (e.g.

set_output(*[, transform])

Set output container.

set_params(**params)

Set the parameters of this estimator.

set_transform_request(*[, return_norm, ...])

Request metadata passed to the transform method.

transform(y[, X, return_norm, target_scale])

Scale input data.

Attributes

TRANSFORMATIONS

names

Names of determined scales.
fit(y: Series, X: DataFrame)[source]#

Determine scales for each group

Parameters:

  • y (pd.Series) – input data

  • X (pd.DataFrame) – dataframe with columns for each group defined in groups parameter.

Returns:

self

fit_transform(y: Series, X: DataFrame, return_norm: bool = False) → ndarray | Tuple[ndarray, ndarray][source]#

Fit normalizer and scale input data.

Parameters:

  • y (pd.Series) – data to scale

  • X (pd.DataFrame) – dataframe with groups columns

  • return_norm (bool, optional) – If to return . Defaults to False.

Returns:

Scaled data, if return_norm=True, returns also scales

as second element

Return type:

Union[np.ndarray, Tuple[np.ndarray, np.ndarray]]

get_norm(X: DataFrame) → DataFrame[source]#

Get scaling parameters for multiple groups.

Parameters:

X (pd.DataFrame) – dataframe with groups columns

Returns:

dataframe with scaling parameterswhere each row corresponds to the input dataframe

Return type:

pd.DataFrame

get_parameters(groups: Tensor | list | tuple, group_names: List[str] | None = None) → ndarray[source]#

Get fitted scaling parameters for a given group.

Parameters:

  • groups (Union[torch.Tensor, list, tuple]) – group ids for which to get parameters

  • group_names (List[str], optional) – Names of groups corresponding to positions in groups. Defaults to None, i.e. the instance attribute groups.

Returns:

parameters used for scaling

Return type:

np.ndarray

inverse_transform(y: Series, X: DataFrame)[source]#

Rescaling data to original scale - not implemented - call class with target scale instead.

set_transform_request(*, return_norm: bool | None | str = '$UNCHANGED$', target_scale: bool | None | str = '$UNCHANGED$') → GroupNormalizer#

Request metadata passed to the transform method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to transform if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to transform.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

  • return_norm (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_norm parameter in transform.

  • target_scale (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for target_scale parameter in transform.

Returns:

self – The updated object.

Return type:

object

transform(y: Series, X: DataFrame = None, return_norm: bool = False, target_scale: Tensor = None) → ndarray | Tuple[ndarray, ndarray][source]#

Scale input data.

Parameters:

  • y (pd.Series) – data to scale

  • X (pd.DataFrame) – dataframe with groups columns

  • return_norm (bool, optional) – If to return . Defaults to False.

  • target_scale (torch.Tensor) – target scale to use instead of fitted center and scale

Returns:

Scaled data, if return_norm=True, returns also scales

as second element

Return type:

Union[np.ndarray, Tuple[np.ndarray, np.ndarray]]

property names: List[str]#

Names of determined scales.

Returns:

list of names

Return type:

List[str]