plot_covariate_balance: Check covariate balance

In ensley-nexant/cfeval: Causal Forest Evaluation and Visualization

Description

After accounting for propensity, covariate distributions should be balanced between treated and control observations. These functions plot overlaid histograms, one for treated and one for control, of each covariate, using inverse propensity weights.

Usage

plot_covariate_balance_numeric(dat, results, covars = NULL, ...)

plot_covariate_balance_categorical(dat, results, covars = NULL)

Arguments

dat	A data frame of covariates
results	The output from tidy_cf
covars	A character vector of covariates to include in the plot. If omitted, all columns of dat for which is.numeric = TRUE (for the numeric version) or is.numeric = FALSE (for the categorical version) will be plotted.
...	Additional arguments passed to geom_histogram

Value

A ggplot2 plot object, or list of plot objects if plot = FALSE.

Examples

## Not run: 
 require(grf)

 n <- 2000; p <- 10

 X <- matrix(rnorm(n * p), n, p)
 dat <- as.data.frame(X)
 dat$a <- sample(letters[1:3], size = n, replace = T)
 dat$b <- sample(letters[10:17], size = n, replace = T)
 W <- rbinom(n, 1, 0.4 + 0.2 * (X[, 1] > 0))
 Y <- pmax(X[, 1], 0) * W + X[, 2] + pmin(X[, 3], 0) + rnorm(n)

 # fit the model and extract results
 cf <- causal_forest(X, Y, W)
 results <- tidy_cf(cf)

 plot_covariate_balance_numeric(dat, results, bins = 20)
 plot_covariate_balance_categorical(dat, results, plot = T)

## End(Not run)

ensley-nexant/cfeval documentation built on May 20, 2020, 12:34 a.m.