Using data mining for prediction of retail banking customer's churn behaviour

For every retail bank, the control over risks that originate from customer fluctuations is of great importance. These fluctuations can occur in the number of active customers or their level of activity. In recent years, Iran has faced many economic difficulties and it has become even more important for banks to be able to preserve their customers and especially the optimum balance for their accounts. The purpose of this paper gains liquidity on the spectrum of accounts and deposits the most desired service. So, in this paper the standard CRISP-DM method has been used for data mining's road map. In addition, this paper applied the churn prediction model and the enablement of customer categorisation on their churn risk. We demonstrate that customer churn may be affected by two additional important factors, namely customer's age and customer's background. Thus, the younger age range corresponds to the highest rate of shedding. Therefore, the highest amount of share among churners corresponds to customers not older than 32 years of age.