sklearn_transformer

to_2d_array(arr: Union[numpy.ndarray, Sequence[Sequence[Any]]]) → numpy.ndarray[source]
class SkLearnTransformerProtocol(*args, **kwds)[source]

Bases: typing_extensions.Protocol

inverse_transform(arr: Union[numpy.ndarray, Sequence[Sequence[Any]]]) → numpy.ndarray
transform(arr: Union[numpy.ndarray, Sequence[Sequence[Any]]]) → numpy.ndarray
fit(arr: Union[numpy.ndarray, Sequence[Sequence[Any]]])
__init__(*args, **kwargs)
class ManualScaler(*args, **kwds)[source]

Bases: sensai.data_transformation.sklearn_transformer.SkLearnTransformerProtocol

A scaler whose parameters are not learnt from data but manually defined

__init__(centre: Optional[float] = None, scale: Optional[float] = None)
Parameters

  • centre – the value to subtract from all values (if any)

  • scale – the value with which to scale all values (after removing the centre)

fit(arr)
transform(arr: Union[numpy.ndarray, Sequence[Sequence[Any]]]) → numpy.ndarray
inverse_transform(arr: Union[numpy.ndarray, Sequence[Sequence[Any]]]) → numpy.ndarray
class SkLearnTransformerFactoryFactory[source]

Bases: object

static MaxAbsScaler() → Callable[[], sklearn.preprocessing.MaxAbsScaler]
static MinMaxScaler() → Callable[[], sklearn.preprocessing.MinMaxScaler]
static StandardScaler(with_mean=True, with_std=True) → Callable[[], sklearn.preprocessing.StandardScaler]
static RobustScaler(quantile_range=(25, 75), with_scaling=True, with_centering=True) → Callable[[], sklearn.preprocessing.RobustScaler]
Parameters

  • quantile_range – a tuple (a, b) where a and b > a (both in range 0..100) are the percentiles which determine the scaling. Specifically, each value (after centering) is scaled with 1.0/(vb-va) where va and vb are the values corresponding to the percentiles a and b respectively, such that, in the symmetric case where va and vb are equally far from the centre, va will be transformed into -0.5 and vb into 0.5. In a uniformly distributed data set ranging from min to max, the default values of a=25 and b=75 will thus result in min being mapped to -1 and max being mapped to 1.

  • with_scaling – whether to apply scaling based on quantile_range.

  • with_centering – whether to apply centering by subtracting the median.

Returns

a function, which when called without any arguments, produces the respective RobustScaler instance.

static ManualScaler(centre: Optional[float] = None, scale: Optional[float] = None) → Callable[[], sensai.data_transformation.sklearn_transformer.ManualScaler]
Parameters

  • centre – the value to subtract from all values (if any)

  • scale – the value with which to scale all values (after removing the centre)

Returns

a function, which when called without any arguments, produces the respective scaler instance.