Application of K-Means Clustering on Retail Data for Customer Monetary-Frequency Segmentation

Rifqi Rahmatullah (Universitas Sebelas April)

Jurnal Intelek Insan Cendikia
Vol. 3 No. 06 (2026): JUNI 2026

The increasing competition in the retail industry requires companies to understand customer behavior more deeply to optimize marketing strategies. Customer segmentation is an essential approach to identify consumer characteristics and design more targeted strategies. This study aims to apply the K-Means Clustering data mining algorithm to segment customers in retail sales data using the Superstore Dataset. The methodology involves preprocessing 9,800 transaction records into 793 unique customer records by extracting Monetary and Frequency attributes. The Elbow Method was then applied to determine the optimal number of clusters. The results indicate that the optimal number of clusters is K=4. The K-Means algorithm successfully classified customers into four segments, namely superstar customers, loyal customers, potential customers, and at-risk customers. These findings provide data-driven insights that can be utilized by companies to design more effective marketing strategies, enhance customer loyalty, and maximize profitability.