In the prediction problem, we are trying to learn from existing data with the primary goal of predicting the outcome in new cases. The simplicity/interpretability of the model is valuable, but may be of less importance than the accuracy of the predictions