README.md

In ncarchedi/iMatchIt: An interactive wrapper for the MatchIt R package

iMatchIt

This is a very early stage project meant to provide an interactive wrapper for the MatchIt R package. The authors of the original MatchIt software have in no way endorsed the work done here.

An iMatchIt session begins with the user loading their data set, then specifying (interactively) which variables represent treatment, outcomes, and covariates. Six different models are defined with matchit() in the background, using the appropriate variables (based on the user's input). The user is then led through a decision tree. At present, each time a decision is made, a list of methods that have been ruled out and those that remain are displayed to the user. This behavior is optional, but helpful for seeing what's going on. When the end of the decision tree is reached, the user is allowed to select a model from those that remain. That model is then dumped into a variable called mod in the user's workspace for further analysis.

To install iMatchIt directly from GitHub, you'll need the devtools R package. Type the following from the R prompt:

install.packages("devtools")
library(devtools)

Then, to install and run iMatchIt:

install_github("ncarchedi/iMatchIt")
library(iMatchIt)
go()

Thanks to @wilcrofter for designing the digraph (directional graph) structure that underlies the iMatchIt decision tree.

ncarchedi/iMatchIt documentation built on May 23, 2019, 1:05 p.m.