drop_multicollin: Drop Multicollinear Columns

In kbal: Kernel Balancing

Drop Multicollinear Columns

Description

Drops multicollinear columns in order of highest correlation using the correlation matrix. This function uses the cor function from the stats package to calculate the correlations between columns.

Usage

drop_multicollin(allx, printprogress = TRUE)

Arguments

allx	a matrix of data to check for multicollinearity. All columns must be numeric.
printprogress	logical to indicate if progress should be printed out to the command line. Default is TRUE.

Value

A list containing:

allx_noMC	resulting data matrix of full rank after multicollinear columns have been dropped.
dropped_cols	column names of the dropped columns.

Examples

# Create data with multicollinearity 
data <- data.frame(x = rnorm(100),
                   y = sample.int(100, 100), 
                   z = runif(100, 3, 6))
test = data.frame(mc_1 = data$x,
                  mc_2 = data$x * 2 + data$y - data$z)
dat = cbind(test, data)
# Run function
mc_check = drop_multicollin(dat)
mc_check$dropped_cols 

kbal documentation built on April 3, 2025, 6:04 p.m.