combine_variables: Compute Weighted Combined Score from Multiple ariables

In eVCGsampler: VCG Sampling using Energy-Based Covariate Balancing

Compute Weighted Combined Score from Multiple ariables

Description

Creates a single combined variable from multiple ariables, weighted by relative importance scores. Variables are scaled before combination.

Usage

combine_variables(formula, data, weights = NULL)

Arguments

formula	A formula specifying the ariables, e.g., '1 ~ cov1 + cov2 + cov3'. The left side should be 1 (placeholder).
data	A data frame containing the variables specified in the formula.
weights	A numeric vector of relative importance weights for each variable. Must have the same length as the number of variables in the formula. Higher weights indicate greater contribution (correlation) to the combined score. If NULL, equal weights (all 1s) are used. Default: NULL.

Details

Variables are first scaled to mean 0 and standard deviation 1, then multiplied by their importance weights and summed. The final score is a weighted linear combination: score = w1*z1 + w2*z2 + ... + wk*zk, where zi are the scaled variables and wi are the weights.

Value

A numeric vector of the same length as nrow(data), representing the combined weighted score for each observation. The score correlates with each input variable proportionally to its importance weight.

Examples

dat <- data.frame(
  age    = rnorm(80, 5, 2),
  weight = rnorm(80, 11, 2),
  class  = rbinom(80, 3, 0.5)
)

# Equal weights
score <- combine_variables(1 ~ age + weight + class, data = dat)

# Custom weights: age contributes 2x, weight 1.5x, class 1x
score <- combine_variables(1 ~ age + weight + class, data = dat,
                           weights = c(2, 1.5, 1))

eVCGsampler documentation built on March 10, 2026, 5:07 p.m.