PERBANDINGAN KINERJA ALGORITMA DATAMINING UNTUK PREDIKSI KELULUSAN MAHASIWA

Abstract

Along with the development of technology, especially the development of increasingly large data storage. One organization that has large data storage is an educational organization. Educational organizations use data to obtain information, especially information about students. Student data has many attributes so that we can make predictions such as predictions of student performance, predictions of scholarship recipients and predictions of student graduation. Data mining methods in education are classified into five dimensions, one of which is prediction, such as predicting output values based on input data. From the results of the research conducted from the initial stage to the testing stage of the application of the C4.5 Algorithm, the accuracy results are higher than Naïve Bayes because in its classification stage, C4.5 processes attribute data one by one. The difference is with naïve Bayes which is influenced by the amount of data used, the comparison of the amount of training and testing data. The feasibility of the model obtained is supported by the high accuracy, precision, recall and AUC obtained from the two algorithms that have been tested. The C4.5 algorithm has an accuracy rate of 79.91%, 89.06% precision and 81.38% recall and an AUC value of 0.823. Meanwhile, Naïve Bayes has an accuracy rate of 76.95%, precision of 75.95% and recall of 98.38% and an AUC value of 0.838.