Forecasting students´ future academic records using Bayesian network

Most educational institutions nowadays digitally manage their students´ data, which are maintained on servers on campus. We intend to use these data in order to provide academic guidance. Thus, in this study, we forecast students´ future academic records using smart card based time recording data and grade data. We use a Bayesian network as our forecasting method. We accordingly revised the data and the variables when designing the Bayesian network forecasting model. We also increased the amount of data and modified the method used to discretize the objective variable, which improved forecasting accuracy.