•  
  •  
 

Author ORCID Identifier

Mohammed Jibril: https://orcid.org/0009-0003-3597-151X

Abstract

Machine learning (ML) has emerged as a powerful tool in education that enhances data-driven decision-making and supports timely interventions. However, its application in Nigerian higher education remains underexplored, particularly where sociocultural and institutional factors may influence predictive accuracy. This study employs the CRISP-DM framework to address this gap by applying ML to predict student performance in core educational technology courses. The models were trained on a dataset of 19,961 instances collected from three Nigerian universities from 2010 to 2024. A percentage split of 80% for training and 20% for testing was adopted. Thirteen machine learning algorithms were applied to predict students’ final exam scores, which included Hist Gradient Boosting Regressor, XGBoost Regressor, and Neural Network Regressor. The Hist Gradient Boosting Regressor emerged as the best model. It achieves a Mean Absolute Error (MAE) of 7.271, Root Mean Squared Error (RMSE) of 9.309, and Median Absolute Error of 6.049. XGBoost closely followed with an MAE of 7.729 and RMSE of 9.540. The best model has been deployed on Streamlit, which is accessible at eduscore.streamlit.app, allowing educators to predict student academic performance and provide timely interventions. The study concluded that machine learning models, particularly Gradient Boosting, can effectively predict student performance and support early identification of at-risk learners in core educational technology courses. It recommends that faculties of education adopt predictive tools, such as the deployed Streamlit model, to enhance academic support and improve students’ performance.

First Page

20

Last Page

37

Ethics Approval

Yes

Declaration Statement

Data Availability Statement All data utilized in this study are included in the manuscript. Additional datasets generated or analyzed during the research can be made available upon reasonable request from the corresponding author.

Funding Statement This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of Interest Disclosure The authors declare no conflict of interest. There are no financial, professional, or personal relationships that could have influenced the research presented in this manuscript.

Participant Consent Statement Informed consent was obtained from all participants involved in this study.

Permission to Reproduce Materials from Other Sources N/A.

Clinical Trial Registration N/A.

Download

Share

COinS