Package 'StratifiedBalancing'

Title: Stratified Covariate Balancing
Description: Performs Stratified Covariate Balancing with Markov blanket feature selection and use of synthetic cases. See Alemi et al. (2016) <DOI:10.1111/1475-6773.12628>.
Authors: Farrokh Alemi <[email protected]> , Amr ElRafey <[email protected]>
Maintainer: Amr ElRafey <[email protected]>
License: GPL (>= 2)
Version: 0.3.0
Built: 2024-11-28 06:36:50 UTC
Source: CRAN

Help Index


Function Stratify

Description

This is the main function which perfroms Stratified Covariate Balancing. It also enables the user to opt for stratifying only the parents in the Markov Blanket of the Treatment variable along with discretization and use synthetic cases to match the data.

Usage

stratify(Treatment, Outcome, Matrix, Discretize, Synthetic, Ordered , Markov)

Arguments

Treatment

The column number of the Treatment variable

Outcome

The column number of the outcome variable

Matrix

The name of the data.frame or matrix where the data is stored

Discretize

A TRUE/FALSE parameter indicating whether covariates should be discretized or not

Synthetic

A TRUE/FALSE parameter indicating whether synthetic matching should be used or not

Ordered

A TRUE/FALSE parameter indicating whether the variables provided are ordered or not

Markov

A TRUE/FALSE parameter indicating whether Markov Blanket of the treatment should be used for rstratification or not

Examples

#create snthetic data
m=matrix(nrow=100,ncol=5,data=0)
for(i in 1:ncol(m)){
m[,i] = rbinom(100,1,0.5)
}
colnames(m)=c("Var1" , "Var2" , "Var3" , "var4" , "Var5")

## Now use Stratify
g=stratify(4,5,m)