Research Specifications

Forecasting future behavior of customers has significant importance in businesses. Consequently, data mining and prediction tools are increasingly utilized by firms to predict customer behavior and to devise effective marketing programs. When dealing with multiple time series data, we encounter with the problem that how to use those time series to forecast the behavior of all customers more accurately. In this study we proposed a methodology to create customer segments based on past data, create Segment-Wise forecasts and then discover the future behavior of each segment. The proposed methodology utilizes existing data mining and prediction tools including time series clustering and forecasting, but combines them in a unique way that results in higher level models in terms of accuracy than baseline model. The proposed methodology has substantial application in marketing for any firm in any domain where there is a need to forecast future behavior of different customer group in an effective manner.