SRxgboost_select_variables: SRxgboost_select_variables

In samuelreuther/SRxgboost: Helper and meta-functions for modelling with XGBOOST

SRxgboost_select_variables

Description

Take the results of a previous run and deduct relevant variables: - by comparing the information gain against the mean of random_numeric, random_binary and random_category variables created in 'SRxgboost_data_prep' - removing from highly correlated variables (default threshold = 0.9 like the default value in recipes::step_corr()) the variable with lower information gain Requirements of previous run with all variables provided at 'lauf_all_variables': - run the following 3 functions: - 'SRxgboost_data_prep(..., add_random_variables = TRUE)' - 'SRxgboost_run' and - 'SRxgboost_plots' Returns a data.frame with relevant variables and their variable importance

Usage

SRxgboost_select_variables(lauf_all_variables, threshold_cor = 0.9)

Arguments

lauf_all_variables	character
threshold_cor	numeric (default = 0.9)

Value

data.frame with relevant variables and their variable importance

samuelreuther/SRxgboost documentation built on March 30, 2025, 12:48 a.m.