Skripsi
ANALISIS CLUSTER K-MEANS UNTUK MENGELOMPOKKAN PROPINSI DI INDONESIA BERDASARKAN RATA-RATA UPAH BERSIH BURUH PER JENIS PEKERJAAN UTAMA
K-Means is a non-hierarchical clustering algorithm used to group data into several clusters based on the similarity of their characteristics. This study aims to determine the optimal number of clusters and cluster characteristics in grouping provinces in Indonesia based on wages by primary occupation. This study uses 8 variables, namely wages for professional workers, managerial workers, administrative workers, sales workers, service workers, agricultural workers, production workers, and other workers. The provinces were grouped using 2, 3, 4, 5, and 6 clusters, and the optimal number of clusters was evaluated using the Elbow and Silhouette Coefficient methods. The best optimal cluster result using the Elbow method was 3 clusters, and the best optimal cluster result using the Silhouette Coefficient method was 2 clusters.The final selection of three clusters as the optimal number was based on the fact that the average Silhouette score indicated a weak structure. The three cluster grouping revealed that Cluster 1 is characterized by moderate wage levels for the variables Professional Workforce, Leadership, Administration, and Sales, and high average wage levels for the other variables. These characteristics encompass 13 provinces, namely Bangka Belitung, Bali, Central Kalimantan, and others. Cluster 2 exhibits characteristics with moderate average wage levels for the variables Leadership Workforce and Other Workforce, and low average wage levels for the remaining variables. These characteristics form the remaining 21 provinces, where wage increases are needed. Cluster 3 is characterized by high average wages, except for the Agricultural Labor and Other Labor variables, which have moderate average wages. It comprises 4 provinces: Kepri, DKI Jakarta, West Java, and Banten. Keywords : K-Means Clustering, Wages, Elbow Method, Silhouette Coefficient
No other version available