Time series predict customer behavior analysis mobile banking

The purpose of this paper is to help businesses use datamining to analyze customers' dynamic behavior in mobilebanking and then drive augmented customer relationships overtime. The tool is used for customer relationships Traditionally,customer segmentation strategies are employed when dealingwith a large community of customers. The research recommendsan innovative methodology for forecasting segment-levelcustomer behavior. This method makes use of both theequipment and methods used by forecasters, such ascomputational intelligence methods and traditional time seriesforecasting methods. An assessment of the suitability of thesuggested method is conducted, as well as a case study based ondata from bank customers' mobile banking. The result of theARMA method in the RMSE evaluation model is ۰.۳۷ and theconfusion matrix has an accuracy of ۰.۸۹ for KNN (K = ۱, lag-۱:۴).

Keywords:

ARMA , ARIMA , KNN , SVMs , Customer behavior , Time series clustering , RFM model Introduction(Heading ۱)