DTRtree: Tree-based Reinforcement Learning for estimating optimal DTR.

In Team-Wang-Lab/T-RL: Tree-based Reinforcement Learning for estimating optimal DTR.

Description

a tree-based reinforcement learning (T-RL) method to directly estimate optimal DTRs in a multi-stage multi-treatment setting. At each stage, T-RL builds an unsupervised decision tree that directly handles the problem of optimization with multiple treatment comparisons, through a purity measure constructed with augmented inverse probability weighted estimators.

Usage

DTRtree(
  Y,
  A,
  H,
  pis.hat = NULL,
  m.method = c("AIPW", "randomForest"),
  mus.reg = NULL,
  depth = 5,
  lambda.pct = 0.05,
  minsplit = 20,
  lookahead = F
)

Arguments

Y	A vector of outcome of interest.
A	A vector of observed treatment options.
H	A matrix of covariates before assigning final treatment, excluding previous treatment variables.
pis.hat	Estimated propensity score matrix.
m.method	Method for calculating estimated conditional mean.
mus.reg	Regression-based conditional mean outcome.
depth	Maximum tree depth.
lambda.pct	Minimal percent change in purity measure for split.
minsplit	Minimal node size.
lookahead	Whether or not to look into a further step of splitting to find the best split.

Team-Wang-Lab/T-RL documentation built on Jan. 3, 2020, 12:11 a.m.