mutlicol_terminator: Multicollinearity TerminatoR - Feature Selection to remove...

In FeatureTerminatoR: Feature Selection Engine to Remove Features with Minimal Predictive Power

Description

This function looks at highly correlated features and allows for a correlation cutoff to be set. Outputs from this function allow for correlations and covariance matrices to be created, alongside visuals and the ability to remove highly correlated features from your statistic pipeline.

Usage

mutlicol_terminator(df, x_cols, y_cols, alter_df = TRUE, cor_sig = 0.9)

Arguments

df	The data frame to pass with the x and y variables
x_cols	The independent variables we want to analyse for multicollinearity
y_cols	The dependent variables(s) in your predictive model
alter_df	Default=TRUE - Determines whether the underlying features are removed from the data frame, with TRUE being the default.
cor_sig	Default=0.9 - A correlation significance for the cut-off in inter-feature correlation

Value

A list containing the outputs highlighted hereunder:

det

• "rfe_model_fit_results" a list of the model fit results. Including the optimal features

• "rfe_reduced_features" a data.frame object with the reduced variables and data

• "rfe_original_data" a data.frame object with the original data passed for manual exclusion based on fit outputs

• "rfe_reduced_data"output of setting the alter_df=TRUE will remove the features / IVs from the data.frame

Examples

library(caret)
library(FeatureTerminatoR)
library(tibble)
library(dplyr)
df <- iris
mc_fit <- mutlicol_terminator(df, 1:4,5, cor_sig = 0.90, alter_df = TRUE)
#View the correlation matrix
mc_fit$corr_matrix
#View the reduced data
head(mc_fit$feature_removed_df,10)

FeatureTerminatoR documentation built on July 14, 2021, 9:06 a.m.