Results are not great; there seems to be a fairly big overprediction at lower wind speeds and corresponding underprediction around rated speed. I am wondering if the categorical variable of the Turbine operating region is causing trouble...
Seems to work now. Added mtry =2 in the call to randomForest in TrainRegressionTree.R so that there is a random choice between more than one variable at any one node when training the regression tree.
Have coded up a randomForest implementation using the Breiman code (http://cran.r-project.org/web/packages/randomForest/index.html).
Results are not great; there seems to be a fairly big overprediction at lower wind speeds and corresponding underprediction around rated speed. I am wondering if the categorical variable of the Turbine operating region is causing trouble...