Prediksi Lama Studi Mahasiswa Menggunakan Algoritma Support Vector Regression (SVR)

Abstract

The issue of on-time graduation in higher education remains a significant challenge, impacting both the quality of education and institutional accreditation. Many students face delays in completing their studies, which affects academic planning and resource management in universities. Therefore, it is essential to develop a predictive model with low error rates and high accuracy to assist educational institutions in designing academic strategies and facilitating timely graduation. This study aims to predict students' length of study using the Support Vector Regression (SVR) algorithm. The dataset used in this research consists of academic records of graduates from the Bachelor’s Program in Computer Science and Information Technology at the University of North Sumatra, covering students admitted between 2016 and 2020. The collected data underwent a series of preprocessing steps, including data cleaning and transformation. After preprocessing, the data was split into training and testing sets. The model was trained using the SVR algorithm, which is well-known for its effectiveness in handling high-dimensional data. The results indicate that the optimized SVR model achieved good predictive performance, indicated by the MAE value of 0.3561, MSE value of 0.2174, MAPE of 7.6190%, and R² of 0.4550.