Analyzing Student Performance Using Classification Algorithms and Association Rule Mining.

Ahmed, banan AbdElnasser; Raslan, Esraa ahmed; Shaheen, Masoud Esmail Masoud

doi:10.21608/ifjsis.2025.343058.1103

	Analyzing Student Performance Using Classification Algorithms and Association Rule Mining.
Labyrinth: Fayoum Journal of Science and Interdisciplinary Studies
Articles in Press, Accepted Manuscript, Available Online from 12 March 2025 PDF (1.25 MB)
Document Type: Original full papers (regular papers)
DOI: 10.21608/ifjsis.2025.343058.1103
View on SCiNiTO
Authors
banan AbdElnasser Ahmed ¹; Esraa ahmed Raslan²; Masoud Esmail Masoud Shaheen³
¹department of computer science faculty of Computers and Artificial intelligence
²Information system department , faculty of computer and Artificial Intelligence Fayoum University
³Department of compute science, Faculty of computers and Artificial intelligence, Fayoum University
Abstract
Predicting student performance is crucial in the educational sector, as analyzing student status can lead to improve performance. Educational data mining is a research field focused on using real-world online data to improve education systems. This data includes academic, socioeconomic, and demographic details for 524 students, encompassing twenty-two attributes. In this study, the Apriori algorithm was employed for association rule mining to conduct an in-depth analysis of student grades and to explore correlations between foundational professional courses and core professional courses. We compared the performance of classification algorithms such as Quest, Random Forest, and Bayes Network Classifiers. Three classification algorithms were implemented using IBM SPSS Modeler, The study highlighted several factors influencing the accuracy of predictions. Random Forest, which achieved the highest accuracy (89%), was particularly sensitive to features like Internal Assessment Percentage (IAP). Quest, with moderate accuracy, emphasized workload-related features such as Theory And Practical Performance (TNP). In contrast, Bayes Network relied on Attendance (ATD) and a diverse range of features, effectively modeling complex interdependencies but at the cost of increased sensitivity to noise. The Apriori algorithm was applied to mine association rules across all attributes, and the most significant rules were displayed.
Keywords
Data mining; Apriori algorithm; Random forest algorithm; Spss modeler; Students’ performance


Statistics Article View: 87 PDF Download: 47