rapid_models.preprocessing.scalers

Convenience scaling functions for rapid-models. For more scaling functions refer to e.g. https://scikit-learn.org/stable/modules/classes.html#module-sklearn.preprocessing

Module Contents

Functions

scale_x_to_box(x, bounds)	Input x = points in [0, 1]^n
scale_x_to_box_inv(x, bounds)	Inverse of scale_x_to_box
standardScaler(x[, mean, std, dim, tensorType, ...])	Standardize features by removing the mean and scaling to unit variance.
standardReScaler(x, mean, std)	Rescale features based on specified mean and std.

rapid_models.preprocessing.scalers.scale_x_to_box(x, bounds)

Input x = points in [0, 1]^n output scaled to lie in the box given by bounds

rapid_models.preprocessing.scalers.scale_x_to_box_inv(x, bounds)

Inverse of scale_x_to_box

rapid_models.preprocessing.scalers.standardScaler(x, mean=None, std=None, dim=0, tensorType='torch', bReturnParam=False)

Standardize features by removing the mean and scaling to unit variance. The standard score of a sample $x$ is calculated as:

$$ x_{out} =

rac{x - x.mean()}{x.std()} $$

Args:

x (array-like, ND): Array of features to be scaled. mean (float, default=None): Specify the mean that will be subtracted from

the features in $x$. If _None_, the mean will be calculated from the features as mean=x.mean().

std (float, default=None): Specify the std that the features in $x$ will

be scaled by. If _None_, the std will be calculated from the features as std=x.std() (unbiased).

dim (int or tuple of python:ints, defalt=0): The dimension or dimensions

to reduce to establish the mean and std if these are _None_.

tensorType (str, default=torch): Specify if torch. or numpy. functions

are used.

bReturnParam (bool, default=False): Specify if the function should return

the mean and std used in the scaling. Should be _True_ if mean and std is _None_ to retain the parameters.

Returns:

x_out (array-like, ND): Scaled features.

• if bReturnParam=True the function return a tuple with (x_out, mean, std)

rapid_models.preprocessing.scalers.standardReScaler(x, mean, std)

Rescale features based on specified mean and std.

$$ x_{out} = x*x.std() + x.mean()

Parameters:

• x (array-like, ND) – Array of features to be rescaled.

• mean (float, default=None) – Specify the mean that will be added to the features in $x$ after rescaling.

• std (float, default=None) – Specify the std that the features in $x$ will be rescaled by.

Returns:

Re-scaled features.

Return type:

x_out (array-like, ND)