Research Article
A Comparative Evaluation of Machine Learning Models for Predicting Student Academic Performance: Baseline Results and Directions for Multimodal Extension
|
International Journal of Computer Applications
Foundation of Computer Science (FCS), NY, USA
|
| Volume 187 - Issue 113 |
| Published: June 2026 |
| Authors: Sajjan Wagle, Purna B. Thapa |
10.5120/ijcaa8ab5d55ed88
|
Sajjan Wagle, Purna B. Thapa . A Comparative Evaluation of Machine Learning Models for Predicting Student Academic Performance: Baseline Results and Directions for Multimodal Extension. International Journal of Computer Applications. 187, 113 (June 2026), 64-67. DOI=10.5120/ijcaa8ab5d55ed88
@article{ 10.5120/ijcaa8ab5d55ed88,
author = { Sajjan Wagle,Purna B. Thapa },
title = { A Comparative Evaluation of Machine Learning Models for Predicting Student Academic Performance: Baseline Results and Directions for Multimodal Extension },
journal = { International Journal of Computer Applications },
year = { 2026 },
volume = { 187 },
number = { 113 },
pages = { 64-67 },
doi = { 10.5120/ijcaa8ab5d55ed88 },
publisher = { Foundation of Computer Science (FCS), NY, USA }
}
%0 Journal Article
%D 2026
%A Sajjan Wagle
%A Purna B. Thapa
%T A Comparative Evaluation of Machine Learning Models for Predicting Student Academic Performance: Baseline Results and Directions for Multimodal Extension%T
%J International Journal of Computer Applications
%V 187
%N 113
%P 64-67
%R 10.5120/ijcaa8ab5d55ed88
%I Foundation of Computer Science (FCS), NY, USA
Abstract
Predicting student academic performance is a critical challenge in educational data mining, with direct implications for early intervention, personalized learning, and institutional resource allocation. This paper presents a systematic comparative evaluation of three widely used machine learning models — Random Forest (RF), Logistic Regression (LR), and Gradient Boosting (GB) applied to a structured student performance dataset comprising demographic information, attendance records, and prior academic scores. Consistent preprocessing pipelines are applied across all three models, including label encoding, mean imputation, feature normalization, and Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance. Results indicate that Gradient Boosting marginally outperforms Logistic Regression in overall accuracy (0.51 vs. 0.50) and precision (0.51 vs. 0.50), while Logistic Regression achieves the highest recall (0.52); Random Forest underperforms both at 0.40 accuracy. The causes of these relatively modest results are analyzed to motivate a proposed extension incorporating CNN-extracted features from handwritten assignment images. This paper contributes a reproducible baseline evaluation framework, a structured analysis of model trade-offs, and a concrete roadmap for multimodal learning analytics research.
References
- S. D. A. Bujang et al., "Multiclass Prediction Model for Student Grade Prediction Using Machine Learning," IEEE Access, vol. 9, pp. 95608-95621, 2021.
- P. Asthana et al., "Prediction of Student's Performance with Learning Coefficients Using Regression Based Machine Learning Models," IEEE Access, vol. 11, pp. 72732-72742, 2023.
- R. Al-Shabandar et al., "Detecting At-Risk Students With Early Interventions Using Machine Learning Techniques," IEEE Access, vol. 7, pp. 149464-149478, 2019.
- Y. Zhang, X. Peng, and T. Huang, "Multimodal Learning Analytics: Combining Trace and Artefact Data for Student Outcome Prediction," Computers & Education, vol. 184, p. 104520, 2022.
- F. Alzubaidi et al., "Towards Unified Deep Learning for Multimodal Medical Data," NPJ Digital Medicine, vol. 6, no. 1, pp. 1-13, 2023.
- T. Chen and C. Guestrin, "XGBoost: A Scalable Tree Boosting System," in Proc. 22nd ACM SIGKDD, 2016, pp. 785-794.
- W. A. Qayyum, "Student Performance Dataset," Kaggle, 2023. [Online]. Available: https://www.kaggle.com/datasets/waqi786/student-performance-dataset
Index Terms
Keywords