Research On Customer Segmentation Model Based On Machine Learning Algorithm

With the progress of the times,enterprises in various industries are paying more and more attention to the importance of customers,and the research based on customer relationship has attracted extensive attention and discussion.In the competitive market,if an enterprise wants to occupy a part of the market,it must need to analyze and study their customers carefully and deeply."Customer-centered" requires enterprises to distinguish the characteristics and needs of different customers,so as to form accurate marketing,which is not only conducive to saving the cost of resources,improve business performance,but also help enterprises to dig into the core needs of customers,explore the direction of reform.On the other hand,with the current upsurge of machine learning,using machine learning to study customer segmentation not only solves the data disaster brought by big data,but also has efficient and scientific theoretical support.Machine learning has unparalleled superiority and can be the booster of customer research.Therefore,combining machine learning to carry out customer research has considerable practical significance and reference value.Based on the real order data of an e-commerce enterprise,the traditional RFM model would be extended in the thesis,five aspects of customer quantity,amount,recent consumption behaviors,frequency and evaluation are comprehensively included to construct a RFATRe model,which is an extended version of RFM model.At the same time,combined with the clustering algorithm in machine learning,the effectiveness of different clustering algorithms is analyzed and compared with the Mini Batch KMeans?AGNES?DBSCAN?CLIQUE?SOM models and algorithms.Based on the idea of integration,the results of multiple algorithm models are synthesized,then the final clustering results are obtained.All customers are divided into four groups and the comprehensive performance of clustering effectiveness is high.According to the four groups,the characters of four groups: high quality customer group,excellent customer group,ordinary customer group and lost customer group are summarized,furthermore,users' value and the profit-making ability could be investigated.