collinear_select: Dual multicollinearity filtering algorithm

In collinear: Automated Multicollinearity Management

Dual multicollinearity filtering algorithm

Description

Automatizes multicollinearity filtering via pairwise correlation and/or variance inflation factors in dataframes with numeric and categorical predictors.

The argument max_cor determines the maximum pairwise correlation allowed in the resulting selection of predictors, while max_vif does the same for variance inflation factors.

The argument preference_order accepts a character vector of predictor names ranked from first to last index, or a dataframe resulting from preference_order(). When two predictors in this vector or dataframe are highly collinear, the one with a lower ranking is removed. This option helps protect predictors of interest. If not provided, predictors are ranked from lower to higher multicollinearity. Please check the sections Variance Inflation Factors, VIF-based Filtering, and Pairwise Correlation Filtering at the end of this help file for further details.

Usage

collinear_select(
  df = NULL,
  response = NULL,
  predictors = NULL,
  preference_order = NULL,
  max_cor = 0.61,
  max_vif = 5,
  quiet = FALSE,
  ...
)

Arguments

df	(required; dataframe, tibble, or sf) A dataframe with responses (optional) and predictors. Must have at least 10 rows for pairwise correlation analysis, and 10 * (length(predictors) - 1) for VIF. Default: NULL.
response	(optional; character or NULL) Name of one response variable in df. Used to exclude columns when predictors is NULL, and to filter preference_order when it is a dataframe and contains several responses. Default: NULL.
predictors	(optional; character vector or NULL) Names of the predictors in df. If NULL, all columns except responses and constant/near-zero-variance columns are used. Default: NULL.
preference_order	(optional; character vector, dataframe from preference_order, or NULL) Prioritizes predictors to preserve.
max_cor	(optional; numeric or NULL) Maximum correlation allowed between pairs of predictors. Valid values are between 0.01 and 0.99, and recommended values are between 0.5 (strict) and 0.9 (permissive). Default: 0.7
max_vif	(optional, numeric or NULL) Maximum Variance Inflation Factor allowed for predictors during multicollinearity filtering. Recommended values are between 2.5 (strict) and 10 (permissive). Default: 5
quiet	(optional; logical) If FALSE, messages are printed. Default: FALSE.
...	(optional) Internal args (e.g. function_name for validate_arg_function_name, a precomputed correlation matrix m, or cross-validation args for preference_order).

Value

character vector: names of selected predictors

Pairwise Correlation Filtering

cor_select computes the global correlation matrix, orders predictors by preference_order or by lower-to-higher summed correlations, and sequentially selects predictors with pairwise correlations below max_cor.

Variance Inflation Factors

VIF for predictor a is computed as 1/(1-R^2), where R^2 is the multiple R-squared from regressing a on the other predictors. Recommended maximums commonly used are 2.5, 5, and 10.

VIF-based Filtering

vif_select ranks numeric predictors (user preference_order if provided, otherwise from lower to higher VIF) and sequentially adds predictors whose VIF against the current selection is below max_vif.

Author(s)

Blas M. Benito, PhD

See Also

Other multicollinearity_filtering: collinear(), cor_select(), step_collinear(), vif_select()

Examples

data(vi_smol)

## OPTIONAL: parallelization setup
## irrelevant when all predictors are numeric
## only worth it for large data with many categoricals
# future::plan(
#   future::multisession,
#   workers = future::availableCores() - 1
# )

## OPTIONAL: progress bar
# progressr::handlers(global = TRUE)

x <- collinear_select(
  df = vi_smol,
  predictors = c(
  "koppen_zone",           #character
  "soil_type",             #factor
  "topo_elevation",        #numeric
  "soil_temperature_mean"  #numeric
    ),
  max_cor = 0.7,
  max_vif = 5
)

x

## OPTIONAL: disable parallelization
#future::plan(future::sequential)

collinear documentation built on Dec. 8, 2025, 5:06 p.m.