How would one look into the study of categorization in terms of divvying up data with different classifications and deciding the best way to go about that?