No abstract

Predicting students' future performance is important for academic stakeholders as the students' success is the objective of the higher educational institutes. The prediction based on past performance and alignment with the curriculum is crucial to support decision-making action effectively for the university with a coherent vertical curriculum. The result of the prediction can be used to intervene and ensure that the student can graduate on time, also preventing the student from dropping out. In this paper, we proposed a methodology for predicting progressively the students' future assessment including feature engineering. Using the ensemble learning techniques, we adapt the existing Ensemble-based Progressive Prediction so it can be applied on students' data that used the Coherent Vertical Curriculum. In this paper, the behavioral data is used instead of domain knowledge-based data. The results show that the algorithm's accuracy has been improved on a real-world student dataset.