This is one of Gaurav Bansal's attempts at the Kaggle Forest Cover Type Prediction contest. It uses a Random Forest prediction with sampsizes controlled. This model correctly predicted 77.7% of the tree cover on the test data.
The model was created in R. The R code is in the file 'forestcover.R'. The training set is in 'train.csv' and the test data is in 'test.csv'. The model output is in 'ForestCoverResult.csv'.
A full description of the contest can be found at https://www.kaggle.com/c/forest-cover-type-prediction