Predicting Students' end-of-term Performances using ML Techniques and Environmental Data

International Journal of Academic Information Systems Research (IJAISR) 7 (10):19-25 (2023)
  Copy   BIBTEX

Abstract

Abstract: This study introduces a machine learning-based model for predicting student performance using a comprehensive dataset derived from educational sources, encompassing 15 key features and comprising 62,631 student samples. Our five-layer neural network demonstrated remarkable performance, achieving an accuracy of 89.14% and an average error of 0.000715, underscoring its effectiveness in predicting student outcomes. Crucially, this research identifies pivotal determinants of student success, including factors such as socio-economic background, prior academic history, study habits, and attendance patterns, shedding light on the nuanced dynamics of student performance. The key influential features identified in this study offer valuable insights into the complex factors shaping student achievement. These insights are vital for educators, policymakers, and institutions seeking to enhance educational outcomes and promote equitable access to quality education. This research provides a data-driven foundation for proactive interventions, personalized learning strategies, and support systems, ultimately contributing to improved student performance and academic success. The high accuracy of the predictive model and the feature analysis it provides empower decision-makers in the education sector. This model holds significant potential for applications in student performance monitoring, early intervention, and tailored educational strategies. By adopting a data-driven approach, this work advances the field of educational analytics and contributes to the goal of fostering student success and educational equity.

Author's Profile

Samy S. Abu-Naser
North Dakota State University (PhD)

Analytics

Added to PP
2023-11-11

Downloads
535 (#42,405)

6 months
138 (#30,275)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?