Penerapan Information Gain Dan Algoritma K-Means Untuk Klasterisasi Kedisiplinan Pegawai Menggunakan Rapidminer

Abstract

One aspect of the discipline of an employee in an agency can be seen from the side of attendance. The level of employee attendance is closely related to an employee's disciplinary assessment. The level of employee discipline can be seen by looking at the hours of attendance or check in attendance, so that with these parameters you will get early, on time and late entry. This study explores data on attendance by using the k-means clustering algorithm. Before calculating the k-means clustering algorithm, attribute selection using information gain is expected to reduce attributes with small weights. The calculations are performed using Rapidminer software. The results showed that the attribute that had the greatest influence was the percentage of late entry with a weight of 0.783. Clustering using the k-means algorithm produces three clusters with the performance value of the Davies Bouldin Index (DBI) -0.645. Cluster zero has fifteen members, cluster one has thirty-six members, and cluster two has fifty-two members. Cluster zero is a cluster that has a low level of discipline, cluster one is a cluster that has a high level of discipline, while cluster two is a cluster that has a moderate level of discipline.