erikcs

causal_survival_forest.custom <- function(
    X, Y, W, D,
    W.hat = NULL,
    target = c("RMST", "survival.probability"),
    horizon = NULL,
    failure.times = NULL,
    num.trees = 2000,
    sample.weights = NULL,
    clusters = NULL,
    equalize.cluster.weights = FALSE,

erikcs

# To use this file include
# source("best_linear_projection_overlap.R")
# after
# library(grf)

best_linear_projection <- function(forest,
                                   A = NULL,
                                   subset = NULL,
                                   debiasing.weights = NULL,
                                   compliance.score = NULL,

erikcs

rm(list = ls())
library(grf)

coeffs = replicate(250, {
  # setup from 
  # https://tompepinsky.com/2022/06/07/illustrating-contamination-bias-with-implications-for-intersectional-description/

  effect1 = 1
  effect2 = 2
  effect3 = 3

erikcs

true = mean(pmax(0.5 * rnorm(1e5), 0))

r = replicate(100, {
  n <- 2000
  p <- 5
  X <- matrix(rnorm(n * p), n, p)
  Z <- runif(n, -4, 4)
  cutoff <- 0
  W <- as.numeric(Z >= cutoff)
  tau <- pmax(0.5 * X[, 1], 0)

erikcs

# GRF with bivariate splits (done in multi_arm_causal_forest) in two specs (separate installs):
# https://github.com/erikcs/grf/commits/mobWY

# a) Using what we inferred to be MOB's "~equivalent" formulation of the GRF CART criterion
# Install with
# devtools::install_github("erikcs/grf", subdir = "r-package/grf", ref = "3ea1ddf") # "mobWY" branch @3ea1ddf Ap="MOB"

# b) Using GRF's criterion
# Install with
# devtools::install_github("erikcs/grf", subdir = "r-package/grf", ref = "c54605a") # "mobWY" branch @c54605a Ap="GRF"

erikcs

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

erikcs

 xtabs(val ~ dgp + estimator + metric + target, DF.table.long, subset = n == 2000)
, , metric = CLF, target = RMST

       estimator
dgp              VT        SRC1        SRC2        IPCW         CSF
  type1  0.26015800  0.23058600  0.25583400  0.21411400  0.19654800
  type2  0.26030400  0.21983600  0.24268800  0.21657200  0.14402200
  type3  0.19990800  0.17275800  0.20015600  0.08673400  0.08790600
  type4  0.19212837  0.13838581  0.15726132  0.03206281  0.02763727

erikcs

### compare with other ways of building a depth 3 tree
library(policytree)
rm(list = ls())
# set.seed(20)
n = 1000
p = 2
d = 3
X = round(matrix(rnorm(n*p),n,p), 2)
Y = matrix(rnorm(n*d), n, d)

erikcs

### compare with other ways of building a depth 3 tree
library(policytree)
rm(list = ls())
# set.seed(20)
n = 1000
p = 2
d = 3
X = round(matrix(rnorm(n*p),n,p), 2)
Y = matrix(rnorm(n*d), n, d)

erikcs

# Master
          oob               test              stdratio         
Y.error   "0.047 +/- 0.001" "0.047 +/- 0.001" "2.471 +/- 0.087"
tau.error "0.009 +/- 0.001" "0.009 +/- 0.001" "1.251 +/- 0.099"
csf.error "0.105 +/- 0.011" "0.104 +/- 0.011" "1.600 +/- 0.201"
mcf.error "0.047 +/- 0.002" "0.048 +/- 0.003" "2.735 +/- 0.143"
                   user.self          sys.self          elapsed           
Y.time             "18.138 +/- 0.057" "0.309 +/- 0.010" "18.496 +/- 0.059"
Y.time.pred        "3.225 +/- 0.039"  "0.107 +/- 0.005" "3.340 +/- 0.042" 
Y.time.ci          "5.135 +/- 0.074"  "0.200 +/- 0.009" "5.348 +/- 0.074"