mutau

gistfile1.txt

set.seed(123)
# DGP with overlay, multi_arm_causal_forest splits on (mu,tau)
r = replicate(50, {
  n=1600
  p=10
  X=matrix(runif(n*p),n,p)
  e = 0.5
  W = rbinom(n, 1, e)
  mu = 2*X[,1] - 1
  TAU = (1 + exp(-20*(X[, 1:2] - 1/3)))^-1
  tau = TAU[,1]*TAU[,2]
  Y = mu + W*tau + rnorm(n)

  cf.default = causal_forest(X, Y, W, W.hat=0.5,num.trees = 500, Y.hat =NULL)
  cf.default.nocent = causal_forest(X, Y, W, W.hat=0.5,num.trees = 500, Y.hat =0)
  cf.mutau = multi_arm_causal_forest(X, Y, as.factor(W), W.hat=c(0.5,0.5),num.trees = 500, Y.hat=NULL)
  cf.mutau.nocent = multi_arm_causal_forest(X, Y, as.factor(W), W.hat=c(0.5,0.5),num.trees = 500, Y.hat=0)

  c(
    mse.default = mean((tau - predict(cf.default)$predictions)^2),
    mse.default.nocent = mean((tau - predict(cf.default.nocent)$predictions)^2),
    mse.mutau = mean((tau - predict(cf.mutau)$predictions)^2),
    mse.mutau.nocent = mean((tau - predict(cf.mutau.nocent)$predictions)^2)
  )

})
summary(t(r))
#  mse.default      mse.default.nocent   mse.mutau       mse.mutau.nocent
# Min.   :0.01435   Min.   :0.02310    Min.   :0.01299   Min.   :0.008669
# 1st Qu.:0.03075   1st Qu.:0.05713    1st Qu.:0.02894   1st Qu.:0.022302
# Median :0.03845   Median :0.06136    Median :0.03639   Median :0.032301
# Mean   :0.04137   Mean   :0.06452    Mean   :0.04036   Mean   :0.033769
# 3rd Qu.:0.05347   3rd Qu.:0.07518    3rd Qu.:0.05105   3rd Qu.:0.041076
# Max.   :0.09690   Max.   :0.10560    Max.   :0.09473   Max.   :0.083380