Riset Deep Learning Coba Prediksi Mahasiswa yang Berisiko Lulus Terlambat

A graduate student at Universitas Pamulang explains why on-time graduation matters for universities, since it serves as a key performance indicator for quality assurance and accreditation. The article notes that at-risk students are often identified too late, leaving little time for intervention. The author plans a thesis comparing a multilayer perceptron deep learning model with tree-based methods such as Random Forest and XGBoost. The piece reviews three 2026 journal articles, including one that found XGBoost outperformed Random Forest on a dataset of 2,456 students, while cautioning that its results may be optimistic. A further study examines how combining academic and non-academic features and different optimizers affects neural network prediction accuracy.