Font Size: 

Students' performance is a very important indicator of the success of an educational institution. For universities in Indonesia, this is measured by the students' Indeks Prestasi Kumulatif (IPK). This research aim to predict students' IPK using Multilayer Perceptron (MLP) Neural Network, with the students' Indeks Prestasi Semester (IPS) from their first semester as the training data (initial values). Data from over 200 of students in the Department of Natural Science Education are collected. The data are then prepared, preprocessed and normalized. Then, the data are split, 80% for training and 20% for validation (testing). After training, the MLP neural network model used are evaluated and its accuracy are calculated. With the variation of the number of hidden node and the weight optimization algorithms, the optimum model are chosen to represent the best predictor of students' performance.

student's performance; prediction; neural network