PREDIKSI LULUS TEPAT DAN TIDAK TEPAT WAKTU MAHASISWA MENGGUNAKAN ALGORITMA K-MEANS
Teknik Informatika STT Nurul Jadid Probolinggo
Universities are educational units being the last chance for someone who want to pursue knowledge through formal education. The number of students who graduate on time becomes an indicator of success of public and private universities. Many studies on predicting students’ on-time graduation have been conducted. In this study, data mining predictive methods was conducted, namely k-Means algorithm which was applied to the data of students who graduated on time and late. K-means algorithm is one of the algorithms in the Data Mining methods. Based on the results of measuring the performance of the algorithm by using cross validation, confusion matrix and ROC curve test methods, it was found that the k-Means had an accuracy value of 84.43% and the value was lower than the neural network method value (90.41%). The AUC value for the neural network method showed the highest value of 0.946 and the value of the k-means method was 0.500. However, there were time differences when testing, which were 1 second for k-means algorithm and 10 seconds for neural network method.
Keywords : Prediction of students’ on-time and late graduation, Data Mining, K-Means algorithm,and Neural Network