Developing a Predictive Model to Predict the Academic Performance of Public High School Students: in the case of Addis Ketema general secondary school

Abstract

Academic performance prediction has become an important research area in educational data mining due to its potential to support early intervention and informed decision-making in education systems. In Ethiopia, secondary school performance plays a critical role in determining students’ access to higher education and future career paths. However, limited studies have applied data mining techniques to predict academic performance at the high school level. This study aims to develop a predictive model to analyze and predict the academic performance of public high school students, using Addis Ketema General Secondary School as a case study. The study adopts the CRISP-DM (Cross-Industry Standard Process for Data Mining) methodology to guide the data mining process, including business understanding, data understanding, data preparation, modeling, evaluation, and deployment. Student academic data were collected from four public high schools in Addis Ketema Sub City for the 2021/2022 academic year. The dataset includes students’ academic results, attendance records, subject performance, and related attributes. WEKA data mining software was used to implement classification algorithms such as Decision Tree (J48), Naïve Bayes, and rule-based classifiers to build predictive models. The experimental results indicate that Decision Tree (J48) achieved relatively better performance compared to other algorithms; however, the overall predictive accuracy of the models was moderate. Factors such as data imbalance, missing attributes, and limited inclusion of socio-economic and behavioral variables affected model performance. Despite these limitations, the study demonstrates the usefulness of data mining techniques in identifying patterns related to student academic performance and in supporting early identification of at-risk students. The findings of this study provide valuable insights for educators, school administrators, and policymakers by highlighting the importance of data-driven decision making in secondary education. The study recommends improving data quality, incorporating additional relevant attributes, and applying advanced machine learning techniques in future research to enhance prediction accuracy and support educational planning and intervention strategies.