<![CDATA[Every university faces students who leave without notice, including those who fail to complete their studies and are declared as dropouts (DO). An initial step in addressing student dropout issues can be undertaken using classification techniques. This study aims to classify dropout students using logistic regression, which is compared with the Artificial Neural Network (ANN) method to categorize data into five classifications: Active, Graduated, Potential to Graduate, Potential DO, and DO. The dataset consists of academic records of undergraduate students from the Computer Science program, obtained from PUSTIPADA at UIN Sumatera Utara. The data includes entry year, study duration, semester GPA, cumulative GPA, credits per semester, total credits, and tuition fees. A total of 1,337 student records were divided into 80% training and 20% testing sets. The logistic regression model achieved an accuracy of 93% on the test data, while the ANN model performed better with an accuracy of 96%. This indicates that ANN is more effective in capturing complex and variable patterns in student data. The findings of this study contribute to academic institutions and educational policymakers, particularly in the Computer Science program, by providing insights for decision-making and developing intervention programs to prevent potential dropouts among students with similar characteristics to those in the dataset.]]>
Copyrights © 2025
