Customer Segmentation With Machine Learning for Online Retail Industry

Abstract

In this study, a comparative analysis of various techniques is presented on customer segmentation methods based on online retail data. A few unsupervised machine learning (ML) clustering models such as K-means clustering model, hierarchical clustering model, Density-based Spatial Clustering of Applications with Noise (DBSCAN) model and a traditional model based on recency, frequency and monetary (RFM) clustering are evaluated in terms of the insight each model offers. The traditional model is included in the analysis since clustering models are not optimization models and the goodness of unsupervised models could only be evaluated with a practical business approach. The results are shared, and each model is assessed in terms of usability for marketing and communication strategies. At the end, the strengths and weaknesses of each model are discussed, and a methodology is proposed for selecting the best clustering method when facing the customer segmentation problem. A detailed literature review is also presented covering the developments in the field of artificial intelligence, clustering models in ML and examples of customer segmentation implementations in various industries.