RFM model customer segmentation based on hierarchical approach using FCA

Abstract

Nowadays, every business focuses on customer relationship management (CRM) to deliver their customers better services and to establish a competitive advantage over their competitors. Significantly, customer insights with solid customer relationships improve customer retention and satisfaction, thereby contributing to profit. Thus, customer segmentation based on cluster analysis is critical to customer identification in CRM. In addition, it can identify the potential customers and their needs to be matched with marketing strategies. However, unfortunately, this approach has led to a gap between the marketing persons who care about the business implications and clustering output with the data science complexity barrier. Moreover, most clustering methodologies give only groups or segments, such that customers of each group have similar features without customer data relevance. Thus, this work sought to address these concerns by using a hierarchical approach. This research proposes a new effective clustering algorithm by combining Recency, Frequency, and Monetary (RFM) model with formal concept analysis (FCA). This new methodology uses the advantages of FCA in building the knowledge representation; therefore, the obtained construction contains both implicit and explicit knowledge. Explicit knowledge shows information represented in the hierarchical structure model, while implicit knowledge is embedded in the structure with its implication properties. Thus, the knowledge structure from FCA reveals relationships among data points in an easily understood manner. The proposed model was evaluated and compared with K-means clustering and hierarchical clustering using the online retail II dataset from the UCI Machine Learning Repository. The proposed method provides enough and appropriate information for marketers to perceive the value of the clustering results for creating practical marketing strategies in real-world business by offering the marketers both customer segmentation and the relationships in customer data at the same time.