Advanced Machine Learning Segmentation Model for Behavioral Prediction in Retail Customers

This study proposes an advanced segmentation model based on Machine Learning (ML) to predict customer behavior and loyalty in the retail sector. The research adopts an applied and quantitative approach, using two years of transactional data to optimize segmentation through an extension of the traditional RFM model (Recency, Frequency, and Monetary value). The methodological innovation lies in the incorporation of a new behavioral variable, derived from predictive analysis of customer engagement, which reflects recent interaction with the brand across multiple digital channels. This additional variable enhances the explanatory power of the model by capturing behavioral dimensions not covered by the classic RFM framework. Logarithmic and Box-Cox transformations were applied to correct data skewness, achieving near-normal distributions for the original variables. The results, validated using ANOVA tests, demonstrate statistically significant differences among the generated clusters. The enhanced model identifies high-value customer segments more accurately and anticipates purchase and churn patterns, resulting in more effective and personalized marketing strategies. In conclusion, the inclusion of a complementary behavioral variable strengthens the predictive capacity of the RFM model and reinforces its applicability as a key analytical tool for customer relationship management in highly competitive retail environments.