Learn R Programming

MultiObjMatch (version 1.0.0)

Multi-Objective Matching Algorithm

Description

Matching algorithm based on network-flow structure. Users are able to modify the emphasis on three different optimization goals: two different distance measures and the number of treated units left unmatched. The method is proposed by Pimentel and Kelz (2019) . The 'rrelaxiv' package, which provides an alternative solver for the underlying network flow problems, carries an academic license and is not available on CRAN, but may be downloaded from Github at .

Copy Link

Version

Install

install.packages('MultiObjMatch')

Monthly Downloads

167

Version

1.0.0

License

MIT + file LICENSE

Maintainer

Shichao Han

Last Published

July 4th, 2024

Functions in MultiObjMatch (1.0.0)

An internal helper function that translates the matching index in the sorted data frame to the original dataframe's row index

Create a skeleton representation of the exclusion edge costs

Create cost skeleton

getPropensityScore

Fit propensity scores using logistic regression.

get_balance_table

Generate balance table

Get unmatched percentage

flattenSkeleton

Turns a skeleton representation of edge costs in a network

Penalty and objective values summary

Marginal imbalance vs. exclusion

An internal helper function that transforms the output from the RELAX algorithm to a data structure that is more interpretable for the output of the main matching function

Extract edges from the network

makeInfinitySparseMatrix

Internal helper to build infinity sparse matrix

Helper function to mask edges

change the distance matrix to cost

An internal helper function that gives the index of matching with a wide range of number of treated units left unmatched

Helper function to combine two sparse distances

Create network flow structure

filter_match_result

Filter match result

Extract the supply nodes from the net

get_pairdist_balance_graph

Total variation imbalance vs. marginal imbalance

matrix2edgelist

Helper function to convert matrix to list

Generate a factor for exact matching.

Generate rho pairs

Create a skeleton representation of the edge costs

get_pairdist_graph

Distance vs. exclusion

Solve the network flow problem - twoDistMatch

Solve the network flow problem - basic version

rho_proposition

Generate penalty coefficient pairs

Get matched dataframe

Optimal tradeoffs among two distances and exclusion

summary.multiObjMatch

Generate numerical summary

Penalty and objective values summary

An internal helper function that translate the matching index in the sorted data frame to the original dataframe's row index

Visualize tradeoffs

Add exclusion edges

Create a skeleton representation of the balance edge costs

Add fine balance edges

An internal helper function that combines two distance object

combine_match_result

Combine two matching result

check_representative

Check the representativeness of matched treated units

Call relax on the network

build.dist.struct_user

An internal helper function that generates the data abstraction for the edge weights of the main network structure using the distance matrix passed by the user.

build.dist.struct

An internal helper function that generates the data abstraction for the edge weights of the main network structure.

compare_matching

Generate covariate balance in different matches

distanceFunctionHelper

Helper function that change input distance matrix

This is a modified version of the function "dummy" from the R package dummies. Original code Copyright (c) 2011 Decision Patterns.

descr.stats_general

Generate summary statistics for matches

Internal helper function that converts axis name to internal variable name

Data precheck: Handle missing data(mean imputation) and remove redundant columns; it also adds an NA column for indicating whether it's missing

Optimal tradeoffs among distance, exclusion and marginal imbalance

Summarize covariate balance table

Change the edgelist to the infinity sparse matrix