Mining Relations between Courses and Research Directions from Educational Data

Analyzing educational data could provide information of students´ behaviors, based on which education policies would be made properly. Existing data mining methods have been widely applied in traditional educational data analysis. However, methods for major research direction identification and relation analysis between courses and research directions have not been studied yet. Therefore, in this paper, we propose two clustering methods HSminer and QSminer to identify major research directions as well as mine relations between courses and research directions from educational data. These two algorithms both adopt the Latent Dirichlet Allocation (LDA) to mine major research directions with elective courses, and then draw correlations between elective courses with compulsory courses. Experimental results on real educational data show the effectiveness of the two methods.