Data Driven Ecommerce App Suite

Abstract

Electronic Commerce is process of doing business through computer networks. A person sitting on his chair in front of a computer can access all the facilities of the Internet to buy or sell the products.On-line shopping has become the trend and people are more comfortable to buy stu s on-line instead of going to shop.This has increased the competition among di erent store owners to show more relevant products to each user in order to make customers life easy by providing recommendation of certain products which he seeks. Recommender system is one of the applications to predict rating or preference for the items that have not been seen by a user. This system typically produces a list of recommendations. Recommending books, CDs, and other products at amazon.com, news etc.. are examples of such applications to name a few. However, despite these developments, the current generation of recommender systems still requires further improvements to make recommendation methods more accurate and applicable to an even broader range to make customer buying process as simple as ever . Hence, advanced recommendation modelling methods, incorporation of various contextual information into the recommendation process, and measures to determine performance of recommender systems are considered. The rapid growth of the market in every sector is leading to a bigger subscriber base for service providers. More competitors, new and innovative business models and better services are increasing the cost of customer acquisition. In this environment service providers have realized the importance of the retention of existing customers. Therefore, providers are forced to put more e orts for prediction and prevention of churn. In this dissertation,we are focusing on two essential tools for an E-commerce store: Recommender System,Churn detection and prevention model.We have proposed an Item based Recommender System which will recommend viewed also viewed products by considering your current interest only and discarding previous history.We have also proposed a churn detection model which is backed by random forest in order to detect the root cause of customer churn.