In ygeunkim/propensityml: Propensity score weighting using machine learning methods

knitr::opts_chunk$set(
  comment = "#>",
  collapse = TRUE,
  out.width = "70%",
  fig.align = "center",
  fig.width = 6,
  fig.asp = .618,
  fig.pos = "H"
  )
options(digits = 3)

propensityml

Overview

This is an R package to help the SKKU modern statistical methods project. It is basically based on the paper

Lee, B. K., Lessler, J., & Stuart, E. A. (2010). Improving propensity score weighting using machine learning. Statistics in Medicine, 29(3), 337–346. doi:10.1002/sim.3782

Installation

# install.packages("remotes")
remotes::install_github("ygeunkim/propensityml")

Usage

propensityml package aims at estimating propensity score with machine learning methods as in the paper mentioned above.

library(propensityml)

The package provides simulation function that generates the dataset in the paper:

Setoguchi, S., Schneeweiss, S., Brookhart, M. A., Glynn, R. J., & Cook, E. F. (2008). Evaluating uses of data mining techniques in propensity score estimation: a simulation study. Pharmacoepidemiology and Drug Safety, 17(6), 546–555 https://doi.org/10.1002/pds.1555

and additional toy datasets. Consider simulation.

The most simplest scenario, i.e. additivity and linearity model:

(x <- sim_outcome(1000, covmat = build_covariate()))

(fit_rf <- 
  x %>% 
  ps_rf(exposure ~ . - y - exposure_prob, data = .))

We have defined the class named propmod for some usage.

class(fit_rf)

Estimating propensity score:

estimate_ps(fit_rf) %>% head()

ygeunkim/propensityml documentation built on Jan. 1, 2021, 1:44 p.m.