The Estimation of Mixture Customer Interpurchase Length

This research provides a customer interpurchase model for a wide application in marketing management. In this model, the author proposes the multi-interpurchase periodes as mixture exponential distribution and the proportion of different type interpurchases "P" is followed Dirichlet prior distribution which is based on Bayesian Theory. This model can descript mixture heterogeneous customer segments. In this topic, the customers are belongs to different groups but their interpurchase data are mixed in the same database. The mixture interpurchase model can conduct the estimation in the macro view and also can be extended to explore customer hazard rate and the variables which influence the alive probability of customer. Finally, we can calculate the customer value for companies to make strategy decision.