R/polymatching-package.R

In polymatching: A Matching Algorithm for Designs with Multiple Groups

#' Polymatching: Matching in Designs with Multiple Treatment Groups
#'
#' The package implements the conditionally optimal matching algorithm, which can be used to generate
#' matched samples in designs with multiple treatment groups.
#'
#' Currently, the algorithm can be applied to datasets with up to 10 groups and generates
#' matched samples with one subject per group. The package provides functions to generate the matched
#' sample and to evaluate the balance in key covariates.
#'
#' @section Generating the Matched Sample:
#'
#' The function implementing the matching algorithm is \code{\link{polymatch}}. The algorithm is iterative and
#' needs a matched sample with one subject per group as starting point. This matched sample can be
#' automatically generated by \code{\link{polymatch}} or can be provided by the user. The algorithm iteratively
#' explores possible reductions in the total distance of the matched sample.
#'
#' @section Evaluating Balance in Covariates:
#'
#' Balance in key covariates can be evaluated with the function \code{\link{balance}}. Given a
#' matched sample and a set of covariates of interest, the function computes
#' the standardized differences and the ratio of the variances for each pair of treatment groups
#' in the study design. For 3, 4, 5 and 6 groups, there are
#' 3, 6, 10 and 15 pairs of groups and the balance is evaluated before and after matching.
#' The result of \code{\link{balance}} can be graphically represented with \code{\link{plotBalance}}.
#'
#' @name polymatching
#' @importFrom magrittr %>%
#' @importFrom rlang .data
"_PACKAGE"