Implementation of Data Mining for Sales Data Clustering Using K-Medoids Algorithm

Abstract

This study aims to assist a trinket shop in achieving its monthly sales targets by applying data mining techniques using the K-Medoids clustering method. The research was conducted in six main stages: (1) data collection, (2) data cleaning, (3) data mining implementation, (4) evaluation of clustering results using the Davies-Bouldin Index (DBI), (5) determination of the optimal number of clusters (best k), and (6) visualization of clustering results. The data used consists of three selected attributes out of six available attributes. The clustering process with the K-Medoids method produced varying clusters due to the random selection of centroids. Based on the DBI evaluation, the optimal number of clusters was found to be k=3, providing the best clustering results to support the shop's marketing strategies.