IMPLEMENTATION OF THE MODIFIED K-NEAREST NEIGHBOR METHOD FOR CLASSIFICATION OF MAJORS CONCENTRATION SELECTION IN LEARNING

Abstract

Department of Communication studies UIN SUSKA Riau, Indonesia currently applies the selected of interest (concentration) to students aimed at directed stu-dents to more focus on certain courses based on their interests and academic abili-ties. The concentration that exist in the Department of Communication stuides are public relations, broadcasting and journalistic. Determination of concentration se-lection is carried out by the head of the department. However, to select the right concentration for students, Department have several problems in selection the concentration process. This is because students’ data and value of courses must be classified first, then by conducting an interview to determine the level of stu-dent interest. So that with this process requires a long time. For this reason, a system that can help the selection process of concentration is needed. In the datamining, there is a classification to allow users to group data based on their class. In this study classification will used the Modified K-Nearest Neighbor (MKNN) algorithm. In this study, the parameters used are the values of 6 courses namely introduction to public relations, introduction to journalism, the basics of broadcasting, news writing, mass communication and advertising and the interest of the students concerned. The test was carried out on a concentration classifica-tion system that had been built using confusion matrix. Confusion matrix results obtained are the highest accuracy in the 90% data training scenario and 10% data test with k = 2 and k = 3 values that is equal to 78.08%.