Automated Market Analysis by RFMx Encoding Based Customer Segmentation using Initial Centroid Selection Optimized K-means Clustering Algorithm

Abstract

Market analysis including customer segmentation is one of the most important approaches utilized by business owners to analyze customer behavior. Such analysis can provide significant insights and decision support for businesses. Multiple research effort was conducted for market analysis including the Recency, Frequency and Monetary analysis (RFM) in addition to many variations including RFD, RFE, RFM-I and RFMTC. In this research a methodology is proposed to utilize the intermediate vector representation of the introduced RFMx for machine learning toward high precision automatic customer segmentation. In this methodology there’s no need to calculate the actual final RFMx score. The RFMx technique introduces a multimonetary model where each monetary value is assigned different weight to suite the business targets of business owners. The proposed model allowed for finely tuned market analyses on product type or service type level. The results showed significant clustering results that lead to automatic customer segmentation without the need to calculate the final RFMx score.