subgroup_tree: subgroup detection tree

In molson2/subgroupTree: Subgroup Detection Trees with Inference

Description

Fits a tree designed to aggressively seek out subsets of the data with large (or small) average treatment effects

Usage

subgroup_tree(response, treated, X, direction = c("max", "min"), ...)

Arguments

response	numeric outcome of interest
treated	boolean vector of treatment assignments
X	data.frame of predictors; must be either numeric or factor types
direction	"max" to search for largest treatment effect, "min" to search for smallest treatment effect
...	additional arguments to rpart.control, such as maxdepth, etc.

Examples

## Not run: 

set.seed(123)
n = 500
p = 5
treated = sample(c(TRUE, FALSE), n, replace = TRUE)
X = as.data.frame(matrix(rnorm(n * p), n))
high_ate = X[, 1] > 0.5
response = rbinom(n, 1, ifelse(high_ate & treated, 0.9, 0.5))
max_tree = subgroup_tree(response, treated, X, 'max', maxdepth = 2,
                        minbucket = 50)
print(max_tree)

## End(Not run)

molson2/subgroupTree documentation built on Dec. 3, 2019, 10:04 p.m.