Prospects identification scheme for supermarkets by classification of customer behavior using time based analysis of transactional data

Abstract

Prospecting means "finding new customers". It demands effort and resources. Traditional prospecting techniques involved surveying. During surveys raw data is generated which must be cleaned before usage. This cleaning is an overhead. Other techniques require customer demographics. Unfortunately supermarket databases lack customer demographics and acquiring these demographics by any mean is very expensive. So these traditional techniques cannot be used to perform prospecting for supermarkets. Hence our aim is to devise a way that will assist supermarket analysts in identifying future prospects by analyzing the semantics hidden in the customer's past transactional records without using any customer demographics and without conducting any surveys. We will classify the customers on the basis of date and time and then identify the prospective customer classes.