عنوان مقاله [English]
In the past years, researchers considered the proceeds from selling items or services as the most important source of corporate profits, because there was not much competition among companies. Nowadays customers are the most important source of revenue in the business institutions and service companies. Thereupon, customer satisfaction must be plan by company managers in order to preserve current customer and develop new customer in today's competitive conditions. However forecasting the future manner of customer can be useful to allocate budget and limited resource for preservation of the most profitable customers that will do a great help to the managers in order to gain market and increase profitability. In this paper, we present the approach to determinate current customer lifetime value and introduce the developed model to predict the future of customer lifetime value. At the first, the current lifetime value of the customers is determined based on developed RFM and using hierarchy weight method. Then in order to model the downfall probability of customers based on the geometric probability distribution for waiting time, customers must be group based on their characteristics by clustering approach. In this research, it is compared some clustering criteria for determining the best number of clusters. We are used some technical instruments such as Rapid Miner software for data preprocessing and also such as IBM SPSS and expert Choice for clustering analysis and compared theirs abilities. After that, we are modelled customer behavior via Markov chain procedure. Then customer lifetime value estimated for the future customers. The power of this research is the usage of developed RFM in order to weight customers before grouping. Because of this, the optimum number of clusters can be carefully determined. In order to demonstrate the applicability of this approach, the research used on the insurance company employed as the case study.