Fit a Cubist regression model on StackOverflow data and make predictions in a distributed manner with SparkR