bayesvalidrox.surrogate_models.meta_model.transform_y

bayesvalidrox.surrogate_models.meta_model.transform_y(self, y, b_i=0, trafo_type='')

Apply chosen transformation to model outputs per key. Currently supports no transform, bootstrap and PCA.

Parameters

selfobject

An object of class MetaModel that contains needed parameters.

ydict

Output to transform, should contain arrays for each key.

b_iint

Current bootstrap index. This is used in PCA-transformation to use the same number of PCA components in each bootstrap iteration. The default is 0.

trafo_typestring

The type of transformation to apply. Currently supported are ‘bootstrap’: Bootstrap each key with self.b_indices ‘pca’: Principal Component Analysis. The transformation has to be

available in self.

‘’: No transformation

Returns

y_transformdict

Transformed outputs.