The percentage of students passing a course depends heavily on how early at-risk learners can be identified and supported. This pa-per presents an Explainable Artificial Intelligence (XAI) based hybrid model for predicting student academic results using CatBoost and Logistic Regression. CatBoost, a gradient-boosting algorithm, is used to capture complex, non-linear relationships among academic attributes such as at-tendance, internal marks, study hours and previous GPA, while Logistic Regression contributes a transparent, probability-based decision boundary. The two classifiers are combined through a Soft-Voting ensemble that averages their predicted class probabilities, and SHAP (SHapley Additive exPlanations) is applied to the trained model to explain individual and global feature contributions. The system was evaluated on a dataset of 10,000 student records using a 80:20 train-test split. The proposed ensemble achieved an accuracy of 99.10%, precision of 99.37%, recall of 99.02%, F1-score of 99.19% and ROC-AUC of 99.98%, outperforming the individual Logistic Regression and CatBoost models on the combined set of metrics while retaining model interpretability. SHAP analysis identified Study Hours, Previous GPA and Internal Marks as the most influential predictors of student outcome. The resulting system offers a practical, transparent and computationally inexpensive decision-support tool that institutions can use for early identification of academically at-risk students.

This paper presented an Explainable AI-based hybrid model for student result prediction that combines CatBoost and Lo-gistic Regression through a Soft-Voting ensemble, with SHAP used to explain both global and individual predictions. On a 10,000-record academic dataset, the ensemble achieved 99.10% accuracy, 99.37% precision, 99.02% recall, 99.19% F1-score and 99.98% ROC-AUC, while SHAP analysis identified Study Hours, Previous GPA and Internal Marks as the most influential factors behind student outcomes. The system is computation-ally light, handles categorical and numerical academic attributes without extensive pre-processing, and produces explanations that non-technical academic staff can act on directly.

[1]O. Daramola, O. Emebo, I. Afolabi, and C. Ayo, “Implementation of an intelligent course advisory expert system,” Int. J. Adv. Res. Artif. Intell., vol. 3, no. 5, pp. 6–12, May 2014. [2] I. Khan, A. R. Ahmad, N. Jabeur, and M. N. Mahdi, “An artificial intelligence approach to monitor student performance and devise preventive measures,” Smart Learn. Environ., vol. 8, no. 1, p. 27, Sep. 2021. [3]H. A. Mengash, “Using data mining techniques to predict student performance to support decision making in university admission systems,” IEEE Access, vol. 8, pp. 55462–55470, 2020. [4]L. M. A. Zohair, “Prediction of student’s performance by mod-elling small dataset size,” Int. J. Educ. Technol. Higher Educ., vol. 16, no. 1, p. 18, Aug. 2019. [5]G. Akçapınar, A. Altun, and P. As¸kar, “Using learning analytics to develop early-warning system for at-risk students,” Int. J. Educ. Technol. Higher Educ., vol. 16, no. 1, p. 40, Oct. 2019. [6]S. Limanto, J. L. Buliali, and A. Saikhu, “GLoW SMOTE-D: Oversampling technique to improve prediction model perfor-mance of students failure in courses,” IEEE Access, vol. 12, pp. 8889–8901, 2024. [7]X. Huang and J. Li, “Explainable AI models for academic perfor-mance prediction in higher education,” IEEE Trans. Learn. Tech-nol., 2025. [8]N. Omar and R. Hassan, “Student performance prediction using machine learning and explainable AI,” Artif. Intell. Rev., Springer, 2025. [9]A. Kumar and R. Sharma, “Student result prediction using logistic regression,” in Proc. IEEE Conf. Educational Data Mining, 2024. [10]V. Reddy and S. Patel, “Student performance prediction using KNN and SVM,” Lecture Notes in Computer Science (LNCS), Springer, 2025. [11]P. Singh and N. Verma, “Decision tree based academic perfor-mance prediction,” IEEE Access, 2024. [12]M. Gupta and A. Jain, “Random forest model for student perfor-mance evaluation,” Computers & Education, Elsevier, 2024. [13]A. Kord, A. Aboelfetouh, and S. M. Shohieb, “Academic course planning recommendation and students’ performance prediction based on educational data mining,” J. Comput. Higher Educ., Springer, 2025. [14]Y. Ren and X. Yu, “Long-term student performance prediction using learning ability self-adaptive algorithm,” Complex & Intell. Syst., Springer, 2024. [15]S. Khan and T. Ali, “Handling imbalanced student data using SMOTE and machine learning,” in Proc. IEEE Int. Conf. AI in Education, 2024. [16]L. Zhou and Y. Wang, “AutoML based student academic result prediction,” Appl. Artif. Intell., Springer, 2025. [17]H. Li and Q. Chen, “Graph neural networks for learning behav-ior analysis and performance prediction,” Knowledge-Based Syst., Elsevier, 2024. [18]J. Martin and D. Lopez, “Predicting student performance in on-line learning using multidimensional time-series data,” Appl. Sci., MDPI, 2024. [19]Y. Zhang and M. Liu, “Explainable student performance predic-tion using ensemble machine learning,” IEEE Access, 2024. [20]K. Patel and D. Mehta, “Early prediction of at-risk students using machine learning models,” Computers & Education: Artif. Intell., Elsevier, 2024. [21]R. Almeida and P. Costa, “Student academic performance predic-tion using gradient boosting techniques,” Appl. Sci., MDPI, 2024. [22]S. Sharma and P. Kulkarni, “Machine learning approaches for pre-dicting student academic outcomes,” Educ. Inf. Technol., Springer, 2025. [23]S. Verma and R. Gupta, “Hybrid machine learning framework for student result prediction,” J. Appl. Data Sci., Taylor & Francis, 2024. [24]J. Fernandez and A. Morales, “Predicting student success using interpretable machine learning models,” Expert Syst. Appl., Else-vier, 2025. [25]P. Das and S. Nanda, “Educational data mining for early student performance prediction,” J. Educ. Technol. Syst., Springer, 2024. [26]K. Lee and S. Park, “Explainable gradient boosting models for student achievement prediction,” Sensors, MDPI, 2025. [27]P. Mehra and S. Kulkarni, “Student academic result prediction us-ing explainable machine learning models,” IEEE Access, 2024. [28]R. Choudhary and K. Malhotra, “Predicting student success using hybrid classification techniques,” Educ. Inf. Technol., Springer, 2024. [29]M. Ahmed and N. Rahim, “Machine learning-based student per-formance prediction in higher education,” Computers & Educa-tion: Artif. Intell., Elsevier, 2024. [30]J. Park and H. Kim, “Early detection of at-risk students using in-terpretable ML models,” Appl. Sci., MDPI, 2025. [31]A. Joshi and M. Patwardhan, “Student result prediction using en-semble learning and explainable AI,” Artif. Intell. Rev., Springer, 2025. [32]T. Wang and L. Xu, “Academic performance prediction using gra-dient boosting and SHAP explanations,” Expert Syst. Appl., Else-vier, 2024. [33]S. Iyer and P. Rao, “Educational data mining for student perfor-mance evaluation,” in Proc. IEEE Int. Conf. Data Science in Edu-cation, 2024. [34]L. Moreno and D. Sanchez, “Explainable student performance prediction in blended learning environments,” Knowledge-Based Syst., Elsevier, 2025. [35]N. Bansal and V. Arora, “Predicting student outcomes using ma-chine learning and learning analytics,” J. Comput. Higher Educ., Springer, 2024. [36]R. Yadav and S. Mishra, “Hybrid ML models for academic perfor-mance prediction with interpretability,” J. Appl. Data Sci., Taylor & Francis, 2025. [37] F. Khan and Z. Ullah, “Student performance prediction using ex-plainable gradient boosting models,” Sensors, MDPI, 2024. J. Silva and M. Pereira, “A comparative study of machine learning models for student result prediction,” IEEE Trans. Learn. Tech-nol., 2025.

© 2025 International Journal of Engineering and Techniques (IJET).

Digital Object Identifier (DOI)

IJET assigns a unique DOI to every accepted article via Zenodo for persistent linking and citation. Your article’s DOI: https://doi.org/{{doi}}

Open DOI About Zenodo DOI

Close