Customer clustering based on purchased items from social network

Abstract

Customer clustering to extract their purchasing behavior is important for business support and decision making. Due to the growth in e-commerce of social network, huge volume of customer data and items provides precious knowledge about their purchased pattern. This paper proposes an unsupervised learning model to process and perform clustering on data from transaction history of customers; then group them into communities based on certain similar purchased items. In this paper, customer and purchased item data is processed to build a graph representing social network for customers. According to this graph, finding communities is addressed by clustering techniques. Customer segmentation would use customer-purchase transaction data to track buying behavior and create strategic business initiatives. The experimental results on realworld Amazon dataset show the efficiency of the proposed method.