Market segmentation using supervised and unsupervised learning techniques for E-commerce applications

Abstract

Market Segmentation has been a key area of implementation of soft computing techniques in E-commerce applications. Various techniques have been used to achieve maximum results in the classification of the ecommerce market. From stochastic techniques to neural networks, there is a plethora of techniques that have been applied. In this paper, we use self organising Maps (SOMs) an unsupervised learning technique to study the various factors which can be used to segment the market. On the other hand supervised learning techniques such as Nearest Neighbour (NN) and Support vector machine (SVM) are used to quantitatively classify the purchase behaviour based on various factors. The better classification technique is identified through appropriate measures. Further, evolutionary algorithms are used to augment the performance of these classification techniques. Analysis of the results and various factors affecting it is also performed.