2021
Tracking student skills real-time through a continuous-variable dynamic Bayesian network
The field of Knowledge Tracing is focused on predicting the success rate of a student for a given skill. Modern methods like Deep Knowledge Tracing provide accurate estimates given enough data, but beingbased on neural networks they struggle to explain how these estimates are formed. More classical methodslike dynamic Bayesian networks can do this, but they cannot give data on the accuracy of their estimatesand often struggle to incorporate new observations in real-time due to their high computational load.This paper presents a novel method, Performance Distribution Tracing (PDT), in which the distribution of the success rate is traced. It uses a dynamic Bayesian network with continuous random variablesas nodes. By tracing the success rate distribution, there is always data available on the accuracy of any success rate estimation. In addition, it makes it possible to combine data from similar/related skills tocome up with a more informed estimate of success rates. This makes it possible to accurately predictexercise success rates, even when an exercise requires a combination of different skills to solve. Andthrough the use of the right conjugate priors, all distributions are available in analytical form, allowing efficient online updates upon new observations. An application to an engineering physics course has shown that the PDT algorithm works well inreal-life settings. Students mainly appreciate the overview given of their learning process as well as the corresponding practice recommendations.