In using machine learning techniques for predictive tasks, the primary purpose of partitioning a raw data set into training and validation subsets is

Question $2$ options:

to balance the proportion of the target class levels and avoid class$-$imbalance issues

to minimize prediction bias$/$variance

to help the user conduct pre$-$processing tasks such as removing outliers

to minimize the training errors

to help the model choose the most important features