1. Is it necessary to use the same model can't we use other classification model? 2. Can we perform Hyperparameter tuning on the model we selected? 3. Also is it mendatory to size of train/size of cross_val set as mentioned in Notebook or we are free to choose by own