Comparison of KNN and J48 Method in Student Academic Performance Classification
Wisnu Agung Prasetyo, Utomo Pujianto
State University of Malang
Abstract
Students academic abilities differ from one another. There are students who have high academic abilities, so they can take good lectures. However, it is not uncommon to find students with low academic abilities, thus making it difficult for them to take lectures. Furthermore, among students with both categories there are students with average or normal academic ability level. Students with low academic ability level will leave them in college. If this condition continues, it will make them difficult to pass or complete in every course they take. The worst possibility is it will be very difficult for students to finish their education (graduation) and even drop out in the middle of the study. This is where the special handling of the instructor is really needed for them. In this study a comparison is made between two classification methods, namely Decision Tree J48 and k-Nearest Neighbour (KNN). The classification process is done by the Rapid Miner application. The results obtained are the Decision Tree J48 method is better than the KNN. One of the reason is that Decision Tree J48 does better classification on handling large and nominal dataset.
Keywords: Classification, KNN, J48, Decision Tree, Student
Topic: Computer Science and Engineering