The Use of Neural Networks for Predicting Student Academic Performance in Higher Education
This paper investigates the use of neural networks to predict student academic performance in higher education. A hybrid CNN-LSTM model was developed to process temporal data on class attendance and online activity. The dataset included records of 1,200 students with exam scores, attendance, LMS engagement, and demographic data. Preprocessing involved cleaning, one-hot encoding, and Z-score normalization. The model achieved high prediction accuracy (MSE = 0.15, R² = 0.82), outperforming linear regression. For identifying at-risk students, the model reached a ROC-AUC of 0.88. The findings are valuable for the development of early warning systems in higher education institutions that enable the timely identification of students at risk
Сведения
- Журнал
- Trudy Universiteta
- Опубликовано
- 2026 · Т. 1, № 102, С. 459
- Язык
- Английский
- ISSN
- 1609-1825 · 2710-3382
- Издатель
- Karaganda Technical University named after Abylkas Saginov
- Аффилиации
- Akhmet Baitursynuly Kostanay Regional University