Optimizing Group Discussion Generation Using K-Means Clustering And Fair Distribution

Abstract

The development of computer-based learning system today can provide a different learning process in a teaching and learning process, but the problems faced by a teacher is the difficulty in grouping discussion group that has a different value of knowledge and skills, because usually this selection of discussion groups in e-learning is done based on the wishes of each student or randomly regardless of the data of knowledge and skills. This research was conducted with the aim of grouping the discussion groups based on the indicators of knowledge and skill by using k-means clustering analysis at SMK Sore Tulungagung. The knowledge and skills scores of class X students in Pekerjaan Dasar Elektromekanik subjects, The Competence of Electricity Installation Engineering will be used as the basic scores. Then, the students of class X were divided into 2 groups, namely the k-means based group and the random based group for further research. The mean score of knowledge and skills are before the learning process and after the results of the evaluation of the discussion group on the k-means based and the random based group. The k-means based class score increases 4,083 from the average. Before the learning, it was 83.292 and it becomes 87.375 after the evaluation, while the random based class only experienced an increase 0,083 from the average. Before the learning it was 81,250 and it becomes 81,333 after the learning evaluation. Based on the result, grouping the discussion group in a fair way in e-learning on the indicators of knowledge and skills using k-means clustering method shows more visible improvement, so k-means clustering is a more optimal method.