Customer Segmentation: Automatic K-Optimization and RFM-Based K-Means Clustering

In the era of Machine Learning and AI, tasks are supported and integrated to enhance speed, convenience, and accuracy in workflows. One of the most critical application domains is marketing, where the utilization of predictive models has garnered significant attention from experts. In marketing, generating revenue from existing customers typically has lower costs than attracting new ones; hence, segmenting customers to gain a clearer understanding of target audiences and developing tailored marketing strategies is essential. The RFM approach is among the most popular techniques for segmenting customers. Based on these attributes, we can assess the value of customers by analyzing their behaviors: how recently they made a purchase (Recency), how often they buy (Frequency), and how much they spend (Monetary), followed by the application of K-Means clustering. This allows businesses to design specific campaigns for each customer group. However, there remains a lack of detailed documentation on preprocessing steps, data visualization, and model construction in customer segmentation. This study aims to detail a process that a business can follow to segment customers, identify the best number of clusters automatically through the Elbow method—a notable limitation of the K-Means model—and use K-Means to classify customers into corresponding segments. The data utilized in this research comes from the "Online Retail II" dataset, hosted by the UCI Machine Learning Repository, offering businesses a clear and detailed view of the customer segmentation process.