README.md

In rothenhaeusler/tms: Targeted model selection

Targeted selection: parameter-specific model selection

The code to reproduce the figures in the manuscript "Model selection for estimation of causal parameters" can be found in simulations.R.

R-package

This package provides a function for selecting among a set of estimators, if the goal is to minimize the mean-squared error with respect to a finite-dimensional parameter of interest.

How to install

1. The devtools package has to be installed. You can install it using install.packages("devtools").
2. The latest development version can then be installied using devtools::install_github("rothenhaeusler/tms").

Usage

n <- 100
Tr <- rbinom(n,1,.5)
X <- .5*Tr + rnorm(n)
Y <- .5*X +  rnorm(n) + .01*Tr
df <-  as.data.frame(cbind(Tr,X,Y))

surrogate_estimator <- function(df) coef(lm(Y~X,data=df))[2]*coef(lm(X~Tr,data=df))[2]
difference_in_means <- function(df) coef(lm(Y~Tr,data=df))[2] 

# The first argument should be an asymptotically unbiased estimator for the parameter of interest; it serves as a benchmark.
tms(difference_in_means,surrogate_estimator,df)

rothenhaeusler/tms documentation built on June 3, 2022, 10:31 a.m.