Overview of the data ow.The input raw data from two example courses CS130 and MA120 are processed and mapped to a feature vector that covers grades correlation, concept dependency, content overlap, and course complexity relationship between the two courses.The classi er then predicts either positive or negative from these constructed features.