An R package for the conformalized counterfactual estimation with the presence of unobserved confounders.
This R package implements the method for sensitivity analysis of individual treatment effects proposed in our paper: Sensitivity Analysis of Individual Treatment Effects: A Robust Conformal Inference Approach.
cfsens_mgn() conducts sensitivity analysis of individual treatment effects with the marginal guarantee.cfsens_pac() conducts sensitivity analysis of individual treatment effects with the PAC-type guarantee.cfsens_cf_mgn() produces robust predictice intervals for the counterfactuals with the marginal guarantee.cfsens_cf_pac() produces robust predictice intervals for the counterfactuals with the PAC-type guarantee.To install the package, run the following commands in R:
if (!require("devtools")){
install.packages("devtools")
}
devtools::install_github("zhimeir/cfsensitivity")The grf package is required.
We illustrate the usage of the package with synthetic dataset.
## Data generating model
data_model <- function(n, p, Gamma, seed){
set.seed(seed)
beta <- matrix(c(-0.531,0.126,-0.312,0.018,rep(0,p-4)), nrow=p)
X <- matrix(runif(n * p), n, p)
U <- rnorm(n) * abs(1 + 0.5 * sin(2.5 * X[,1]))
Y1 <- X %*% beta + U
Y0 <- X %*% beta - U
prop.x <- exp(X %*% beta) / (1 + exp(X %*% beta))
p.x <- 1-(1/(prop.x + (1-prop.x)/Gamma ) -1)/( 1/(prop.x + (1-prop.x)/Gamma) - 1/(prop.x + Gamma*(1-prop.x)))
t.x <- qnorm(1-p.x/2) * abs(1+0.5*sin(2.5*X[,1]))
prop.xu <- (prop.x/(prop.x+Gamma*(1-prop.x)))*(abs(U)<=t.x) + (prop.x/(prop.x+ (1-prop.x)/Gamma))*(abs(U)>t.x)
TT <- rbinom(n, size=1, prob=prop.xu)
Y <- TT * Y1 + (1 - TT) * Y0
return(list(Y1 = Y1, Y0 = Y0, Y = Y, X = X, T = TT))
}
## Generate confounded training data
n <- 5000
n_test <- 5000
p <- 10
Gamma <- 1.5
data <- data_model(n, p, Gamma, 2021)
## Generate confounded test data
test <- data_model(n_test, p, Gamma, 2022)
## Run sensitivity analysis with marginal guarantee
## grf pacakge needs to be installed
alpha <- 0.2
res <- cfsens_mgn(data$X, data$Y, data$T,
alpha = alpha, null_type = "negative",
ps_fun = regression_forest,
pred_fun = quantile_forest)
out_res <- predict(res, test$X, Y1_test = test$Y1, type = "att")
out_res
## Run sensitivity analysis with PAC-type guarantee
## grf package needs to be installed
alpha <- 0.2
delta <- 0.1
res <- cfsens_pac(data$X, data$Y, data$T,
alpha = alpha, delta = delta, null_type = "negative",
ps_fun = regression_forest,
pred_fun = quantile_forest)
out_res <- predict(res, test$X, Y1_test = test$Y1, type = "att")
out_res
## Construct predictive intervals with marginal guarantees
## grf package needs to be installed
alpha <- 0.2
res <- cfsens_cf_mgn(data$X, data$Y, data$T,
Gamma = Gamma, alpha = alpha,
ps_fun = regression_forest,
pred_fun = quantile_forest)
out_res <- predict(res, test$X, estimand = "treated", type = "ate")
cover <- mean((test$Y1 >= out_res$Y_lo) * (test$Y1 <= out_res$Y_hi))
cover
## Construct predictive intervals with marginal guarantees
## grf package needs to be installed
alpha <- 0.2
delta <- 0.1
res <- cfsens_cf_pac(data$X, data$Y, data$T,
Gamma = Gamma, alpha = alpha,
delta = delta,
ps_fun = regression_forest,
pred_fun = quantile_forest)
out_res <- predict(res, test$X, estimand = "treated", type = "ate")
cover <- mean((test$Y1 >= out_res$Y_lo) * (test$Y1 <= out_res$Y_hi))
cover