COMPARATIVE STUDY OF FUZZY C-MEANS AND K-MEANS ALGORITHM FOR GROUPING CUSTOMER POTENTIAL IN BRAND LIMBACK

Abstract

The customer is a stakeholder for a business, to maintain and increase customer enthusiasm and develop it for the company's performance, it is necessary to do customer segmentation which aims to find out potential customers. This study uses purchase transaction data from Brand Limback customers in the period 2020. The use of RFM (Recency, Frecuency, Monetary) analysis helps in determining the attributes used for customer segmentation. To determine the optimal number of clusters from the RFM dataset, the Elbow method is applied. The datasets generated from RFM are grouped using the Fuzzy C-Means and K-Means algorithms, the two algorithms will compare the quality in the formation of clusters using the Silhoutte Coefficient and Davies-Bouldin Index methods. Customer segmentation from the RFM dataset that has been clustered produces 7 optimal clusters, namely Cluster 0 is a bronze customer. Cluster 1 is a silver customer. Cluster 2 is a gold customer. Cluster 3 is a platinum customer. Cluster 4 is a diamond customer. Cluster 5 is a super customer, and cluster 6 is a superstar customer. The cluster validation of k-means using the silhouette coefficient produces a value of 0.934 while the Davies bouldin index produces a value of 0.155 and the validation results of the fuzzy c-means algorithm using the silhouette coefficient produces a value of 0.921 while the Davies bouldin index produces a value of 0.145.