featureImportance: Calculate the feature importances from a decision ensemble

In leylabmpi/endoR: Interpret and Visualize Stable Tree Ensemble Models

Calculate the feature importances from a decision ensemble

Description

This function calculates the featureImportance of variables in a decision ensemble. Levels of discretized variables are grouped so that only one importance is returned ('variable name' separated by 2*'_' from the 'level name').

Usage

featureImportance(nodes_agg)

Arguments

nodes_agg

a datatable with, for variable in each decision, their name (column Feature), decision-wise importance (column importance), the importance and multiplicity of the decision (columns imp and n).

Value

a datatable with the feature importance for each variable.

leylabmpi/endoR documentation built on Oct. 20, 2023, 10:53 p.m.