PENERAPAN DATA MINING MENGGUNAKAN HIERARCHICAL K-MEANS BERDASARKAN MODEL RFM

Abstract

E-commerce has become an inseparable part of the global retail work environment, where this has an impact by presenting new business competition so that companies are required to determine strategies to be able to gain profits. 
 The purpose of this study is to produce customer segmentation using a combination of Hierachical K-Means Clustering algorithms on online retail transaction data that is transformed into Recency, Frequency, and Monetary (RFM) forms and obtain grouping results that have a high degree of similarity by evaluating clusters that are formed using Silhouette analysis. 
 The results of the study stated that the validation test using the Silhouette Coefficient of the combination of the Hierachical K-Means Clustering algorithm was superior to the K-Means algorithm with the optimal coefficient value of the combination of the K-Means algorithm and the Ward method of Hierachical Clustering, namely 0.54027 with the number of k = 4 while the Hierachical Clustering method was K-Means is only 0.44060 with a total of k = 3. Clustering produces two groups of customers, namely Uncertain and Best Customers according to the customer value matrix.