tabensemb.data.dataprocessor.CorrFeatureSelector#

class tabensemb.data.dataprocessor.CorrFeatureSelector(**kwargs)[source]#

Bases: AbstractFeatureSelector

Select features that are not correlated (in the sense of Pearson correlation). Correlated features will be ranked by SHAP using RandomForestRegressor, and the feature with the highest importance will be selected.

Parameters:

thres:: The threshold of the Pearson correlation coefficient.
n_estimators:: The number of trees used in random forests.

Methods

__init__(**kwargs)#

`_defaults`()	Defaults values for arguments defined in `_cls_required_kwargs()` and `_required_kwargs()`
`_get_feature_names_out`(data, datamodule)	Get selected features.