Finding Synthetic Datasets and Prediction rules which work more suitable for Random Forests.