Research on the course discrimination based on multi-classification method and feature selection

Exam distinction can only reflect a course of the distinction between students. In order to find the main differentiating courses in the university curriculum, the paper proposes the concept of the course differentiation, focusing on the value of course discrimination, the classification method and the proportion of professional courses in selected courses. In order to obtain the value of course differentiation, a method of combining multi-classification and feature selection is proposed. First of all, the data sources of students' achievement are classified by traditional five-level, N-score, M-classification and unsupervised four methods. Then, using the wrapper feature selection method, the classification accuracy rate and the feature subset of each dataset are calculated by different classifiers. Finally, we found the connotation and extension of the course discrimination. Experiments show that the proposed method can find the maximum value of the course distinction and the corresponding classification method, the proportion of professional courses is much larger than the proportion of public courses. It achieves the curriculum and assessment of the distinction requirements.