MaskedMinMaxScaler

class aixd.data.transform.MaskedMinMaxScaler(target_range: tuple = (0, 1), per_column: bool = True, scale_inverse_masked: bool = True)[source]

Bases: MinMaxScaler

Implements min-max scaling of the data with a masked domain. I.e., the masked values are not considered in the min-max computation, and not scaled.

Parameters:

target_range (tuple, default=(0, 1)) – The target range for the scaling.
per_column (bool, default=True) – If set normalization is performed per column for multi-dim DataObjects.
scale_inverse_masked (bool, default=True) – If set, the inverse transformation will scale the masked values, otherwise they are kept as they are. See notes for more details.

Notes

Since masked values are not considered in the min-max computation, more than one value can be mapped to the same value in the target range, making the inverse transformation ambiguous. The argument scale_inverse_masked controls this behavior. If set, the inverse transformation will scale the masked values, otherwise they are kept as they are.

Methods

`inverse_transform`	Inverse transform of the input.
`transform`	Transforms the input.

Inherited Methods

`copy`	Copies the transformation.
`fit`	Fits the transformation.
`fit_transform`	Fits and transforms the input.
`is_fitted`	Returns true if transformation strategy is fitted.
`reset`	Method to implement to reset fitted values.