rf_reg.R

library(randomForest)
library(miscTools)
library(ggplot2)

cols <- c('is_red', 'fixed.acidity', 'density', 'pH', 'alcohol')
rf <- randomForest(alcohol ~ ., data=train[,cols], ntree=20)

(r2 <- rSquared(test$alcohol, test$alcohol - predict(rf, test[,cols])))
# [1] 0.6481
(mse <- mean((test$alcohol - predict(rf, test[,cols]))^2))
# [1] 0.6358

p <- ggplot(aes(x=actual, y=pred),
  data=data.frame(actual=test$alcohol, pred=predict(rf, test[,cols])))
p + geom_point() +
	geom_abline(color="red") +
	ggtitle(paste("RandomForest Regression in R r^2=", r2, sep=""))