A neural network-based approach to a priori assessment of the student academic performance in the Mirera digital learning platform

The paper presents our approach to an a priori assessment of the final student performance in the Mirera digital learning platform. The assessment is based on interim tests at seminars, homework evaluations, and individual tests. In this case, both the test results and the student behavior during the tests are considered. In the proposed approach, students are conditionally divided into three categories: underperforming students with unsatisfactory final results, satisfactory performing students with average results, and high performing students. For each category, the type and feasibility of automating the teacher's corrective actions to improve the student's final scores can be identified. The score is generated using artificial neural networks. The a priori estimate can be used for early detection of underperforming students who need help, as well as for building adaptive learning tracks for average and high performing students. The proposed approach can be applied only to digitally transformed academic process. The authors are implementing adaptive learning technologies in the Mirera digital learning platform.