Prediksi kelulusan tepat waktu mahasiswa untuk pemantauan program studi menggunakan metode data mining

Abstrak

This research will conduct data exploration (data mining) using student data in the undergraduate study program (S1) at PQR University for the 2022/2023 academic year. The study aims to predict students' on-time graduation according to the monitoring requirements of the Accreditation Body (students' timely study period is four years). The test data parameters use student master data, student transaction data, and data on the graduation status of 2019 class students in the 2023/2024 academic year. Data testing and training were conducted using the k-Nearest Neighbors algorithm method. The data training obtained 75% accuracy, 75% precision, and 0% recall value. The data testing results obtained 87.76% accuracy, 89.19% precision, and 83.33% recall value. The data training and data testing results show a high percentage of not passing the monitoring. University leaders can take an early step based on the prediction results to make academic policies to increase the number of on-time graduates.