install.packages("grf")
R.home()
getwd()
setwd("/Users/jocteur/.julia/dev/CausalForest/notebooks")
source("~/.julia/dev/CausalForest/notebooks/R code GRF.R")
setwd("/Users/jocteur/.julia/dev/CausalForest/notebooks")
#data = read.csv("sinus_causal_3.csv", dec=".")
data = read.csv("sinus_causal_4.csv", dec=".")
library("grf")
library("Metrics")
X = data[,1:10]
W = data[,11]
Y = data[,12]
cf = causal_forest(X, Y, W, num.trees = 500)
errors_1 = rep(0, 100)
for (i in 1:100){
set.seed(i)
Xtest = matrix(runif(10000,0,10),nrow=1000)
pred = predict(cf, Xtest)$predictions
true_effect = 25*sin(Xtest[, 1])
errors_1[i] = rmse(true_effect, pred)
}
mean(errors_1)
var(errors_1)
freq = split_frequencies(cf,20)
sum(freq[,1])/sum(freq)
set.seed(1)
Xtest = matrix(runif(10000,0,10),nrow=1000)
pred = predict(cf, Xtest)$predictions
true_effect = 25*sin(Xtest[, 1])
idx <- order(Xtest[, 1])
plot(Xtest[idx, 1], pred[idx], type='l', col='red')
lines(Xtest[idx, 1], true_effect[idx], col="blue")
source("~/.julia/dev/CausalForest/notebooks/R code GRF.R")
data = read.csv("sinus_causal_3.csv", dec=".")
library("grf")
library("Metrics")
X = data[,1:10]
W = data[,11]
Y = data[,12]
cf = causal_forest(X, Y, W, num.trees = 500)
errors_1 = rep(0, 100)
for (i in 1:100){
set.seed(i)
Xtest = matrix(runif(10000,0,10),nrow=1000)
pred = predict(cf, Xtest)$predictions
true_effect = 25*sin(Xtest[, 1])
errors_1[i] = rmse(true_effect, pred)
}
mean(errors_1)
var(errors_1)
freq = split_frequencies(cf,20)
sum(freq[,1])/sum(freq)
set.seed(1)
Xtest = matrix(runif(10000,0,10),nrow=1000)
pred = predict(cf, Xtest)$predictions
true_effect = 25*sin(Xtest[, 1])
idx <- order(Xtest[, 1])
plot(Xtest[idx, 1], pred[idx], type='l', col='red')
lines(Xtest[idx, 1], true_effect[idx], col="blue")
data = read.csv("sinus_causal_3.csv", dec=".")
library("grf")
library("Metrics")
X = data[,1:10]
W = data[,11]
Y = data[,12]
cf = causal_forest(X, Y, W, num.trees = 500)
errors_1 = rep(0, 100)
for (i in 1:100){
set.seed(i)
Xtest = matrix(runif(10000,0,10),nrow=1000)
pred = predict(cf, Xtest)$predictions
true_effect = sin(Xtest[, 1])
errors_1[i] = rmse(true_effect, pred)
}
mean(errors_1)
var(errors_1)
freq = split_frequencies(cf,20)
sum(freq[,1])/sum(freq)
set.seed(1)
Xtest = matrix(runif(10000,0,10),nrow=1000)
pred = predict(cf, Xtest)$predictions
true_effect = sin(Xtest[, 1])
idx <- order(Xtest[, 1])
plot(Xtest[idx, 1], pred[idx], type='l', col='red')
lines(Xtest[idx, 1], true_effect[idx], col="blue")
vec.pac= c("foreign","quantreg", "gbm", "glmnet",
"MASS", "rpart", "nnet", "matrixStats",
"xtable", "readstata13","grf","remotes",
"caret",  "multcomp","cowplot","SuperLearner",
"ranger","reshape2","gridExtra","bartCause","xgboost","bartMachine","nnet")
install.packages(vec.pac, dependencies = TRUE)
