gistfile1.txt

gbmFit.ada = gbm(formula = ylearn ~.,
             distribution = 'adaboost',
             data = xlearn,
             n.trees = 10000,        #the number of trees in the model
             interaction.depth = 5,  #each tree will evaluate five decisions
             n.minobsinnode = 2,     #the number of obs present to yield a terminal node, higher means more conservative fit
             shrinkage = .01,        #the learning rate, dictates how fast the algorithm moves across the loss gradient
             bag.fraction = 0.5,     #subsampling fraction, 0.5 is best
             train.fraction = 0.8,   #fraction of data for training
             cv.folds = 5)           #running five-fold cross-validation

test_data$preds_ada = predict(gbmFit.ada, test_data, n.trees = best.cv.ada, type = 'response')