Integrating RFM and Classification for Response Modeling Based on Customer Lifetime Value

Abstract

One of the most important challenges in direct marketing is finding differences between customers and identifies profitability of each customer for target marketing. Response modeling is an useful technique for this issue that predicts customer ’ s response to a campaign. Accuracy of response model is very important due to high cost and time of marketing process. Due to this, this paper has provided a framework for building an accurate model based on weighted RFM analysis and calculating customer lifetime value (CLV) for each segment of customers, then uses CLV as one of predictor features with demographical features in C5 algorithm. The experimental results show by compacting transactional behaviors of customers in CLV value and using it with demographical features concurrently as predictors of classification algorithm is an efficient method for building response model that is much more accurate than those methods that did not used demographical features and CLV for prediction.