What validation techniques do you use to ensure your predictive models do not overfit the training data?
