In denisagniel/synthate: Synthetic Causal Inference

knitr::opts_chunk$set(
  warning = FALSE, 
  message = FALSE,
  collapse = TRUE,
  cache = TRUE,
  comment = "#>",
  fig.path = "README-"
)

synthate package: Synthetic estimators for the average treatment effect

This package comprises a suite of tools to simulate various data-generating processes. Its purpose is to check the operating characteristics of various causal estimators and to optimally combine them.

Installation

The synthate package can currently be obtained from GitLab: https://code.rand.org/agniel-projects/personal-packages/synthate or from GitHub at https://github.com/denisagniel/synthate. It may be downloaded by running

devtools::install_github('denisagniel/synthate')

Overview

There are three main features of synthate:

Simulating data
Causal estimation
Combining estimators

Simulating data

There are currently seven DGPs derived from six different papers to choose from:

Short-hand | Paper ---------------------|--------------------------------- ks | Kang and Schafer (@Kang2007) ld | Lunceford and Davidian (@Lunceford2004) iw | Waernbaum (@Waernbaum2012) fi | Fan and Imai (@Fan2016) pa | Austin (@Austin2009) ls | Leacy and Stuart (@Leacy2014) ik | Iacus and King (@Iacus2009)