Customer Segmentation and Churn Prediction in Online Retail

Abstract

The online retail industry has changed the way customers shop as everything is available online. In order to build a loyal customer base, a company needs to deploy various marketing strategies focused on the diverse nature of its customers. We propose a model, abbreviated as RFMOC, based on extension of recency frequency, monetary (RFM) analysis with two new variables to segment customers. The model also studies the segmentation performance for the k-means clustering algorithm. Moreover, customer lifetime value (CLV) is calculated for the weighted RFMOC with weights for variables calculated by the analytic hierarchy process (AHP) and customer segments are then ranked accordingly which helps to create targeted marketing strategies. At last, the customer churn prediction is performed using logistic regression by further extending the RFMOC with one more variable, abbreviated as RFMOCD, in order to predict the churning behaviour of the customers. The proposed approach is helpful to assess customer loyalty and to manage customer relationships in an effective manner.