Prediksi Tingkat Kelulusan Menggunakan K-Means Pada Program Studi Informatika Unismuh Makassar

Abstract

Predicting timely graduation brings numerous benefits not only to students but also to the university itself. Creating a graduation prediction model assists students and academic advisors in fostering a positive environment that encourages on-time graduation by developing a predictive model for graduation rates using the K-means data mining method in the Informatics study program at Universitas Muhammadiyah Makassar. This method is used to cluster students based on attributes such as total credits taken, semester Grade Point Average (GPA), and overall Cumulative Grade Point Average (CGPA). The clustering aims to identify patterns and characteristics of student graduation. Data from several semesters is collected and preprocessed, including data normalization and transformation. The research steps involve data preprocessing, cluster labeling, distance calculation to cluster centers, and result analysis. The analysis shows that the K-means method can generate student clusters with varying graduation rate patterns. The formed clusters can be interpreted as groups of students with potential for timely graduation or groups needing more attention to achieve on-time graduation. Empirical validation is performed by comparing K-means prediction results with actual graduation data. Accuracy measurement involves calculating the percentage of similarity between predictions and actual data. Empirical validation results demonstrate the accuracy level, which can serve as a benchmark for assessing the performance of this prediction model. This study aims to provide deeper insights into factors influencing student graduation and potentially support decision-making at the academic level.

Keywords: Graduation Prediction, Data Mining, K-Means, Analysis, Clustering, Empirical Validation.