Scaler

class baybe.scaler.Scaler

Bases: ABC

Abstract base class for all scalers.

Parameters:

searchspace (DataFrame) – The search space that should be scaled.

Public methods

__init__(searchspace)
fit_transform(x, y)	Fit the scaler using the given training data and transform the data.
transform(x)	Scale a given input.
untransform(mean, variance)	Transform mean values and variances back to the original domain.

Public attributes and properties

SUBCLASSES	Class variable for all subclasses
type	Class variable encoding the type of the scaler.

__init__(searchspace: DataFrame)

abstract fit_transform(x: Tensor, y: Tensor)

Fit the scaler using the given training data and transform the data.

Parameters:

• x (Tensor) – The x-data that should be used.

• y (Tensor) – The y-data that should be used.

Return type:

tuple[Tensor, Tensor]

Returns:

The transformed data.

transform(x: Tensor)

Scale a given input.

Parameters:

x (Tensor) – The given input.

Return type:

Tensor

Returns:

The scaled input.

Raises:

RuntimeError – If the scaler is not fitted first.

untransform(mean: Tensor, variance: Tensor)

Transform mean values and variances back to the original domain.

Parameters:

• mean (Tensor) – The given mean values.

• variance (Tensor) – The given variances.

Return type:

tuple[Tensor, Tensor]

Returns:

The “un-transformed” means and variances.

Raises:

RuntimeError – If the scaler object is not fitted first.

SUBCLASSES: dict[str, type[Scaler]] = {'DEFAULT': <class 'baybe.scaler.DefaultScaler'>}

Class variable for all subclasses

type: str

Class variable encoding the type of the scaler.