Applying Deep Knowledge Tracing Model for University Students' Programming Learning

Programming ability has become one of the most critical competencies. In the past, it was hard for teachers to find and understand the problems that students face in the coding process. Nowadays, we could apply data mining methods focused on the educational domain. Moreover, through deep knowledge tracing, we can model students' learning trajectories, understand their current knowledge level, and help students overcome their weaknesses. This study was conducted at a national university in northern Taiwan. A total of 20 graduate students participated in the experiment for 16 weeks. This study combines deep knowledge tracing to develop a program learning system. The system supports the predictions based on the data accumulated from students' learning processes. The system dashboard can immediately help students and teachers understand students' learning behavior and mastery of various knowledge points and provide corresponding learning suggestions. The results show that students' program ability has been significantly improved in this study. Deep knowledge tracing can effectively be used in programming classes to evaluate students' abilities according to their different knowledge points.