Perbandingan Algoritma K-Means dan Hybrid K-Means KNN untuk Pembagian Kelas Kuliah Mahasiswa
Problems that occur when the division of student classes is the difference in the ability of students in each class that can impact on the ineffectiveness of the learning process in the class. Grouping students with the same ability is very important in order to improve the quality of teaching and learning process. By grouping the right students, they will be able to help each other in the learning process. In addition, dividing the class of students according to their ability can facilitate the educator in determining the appropriate method or learning strategy. The use of appropriate methods and learning strategies will improve the effectiveness of the teaching and learning process. This research compared the K-Means and Hybrid K-Means KNN algorithms in division of students classes. The K-means algorithm is used for class division of student lectures based on the assessment component of the pre-requisite course. The features used in the grouping is the value of the task, the value of the midterm exam, the value of the final exam of the semester, and the cumulative achievement index (GPA). The KNN algorithm is used to predict students' graduation in a course based on previous data. This prediction result will be used as an additional feature used in the formation of student classes using K-means algorithm. From the test results can be concluded that the number of clusters or classes and the amount of data used affects the quality of clusters formed by the K-Means and Hybrid K-Means KNN algorithms used. The value of the Silhouette Index of the K-Means KNN algorithm is 0.534 (classified as medium structure), 0.057 higher than the K-Means algorithm. So it can be concluded that the class quality generated by the K-Means KNN hybrid algorithm is better.