CUSTOMER CHURN ANALYSIS WITH BACK-PROPAGATION NEURAL NETWORK: CASE INTERNET SERVICE PROVIDER XYZ

Ali Reza Yudhistira .
2016 International Journal of Research in Engineering and Technology  
Number of Internet users worldwide is likely to increase exponentially. It is an opportunity for companies that specialize in Internet Service Provider (ISP). An ISP in Surabaya, ISP xyz, their customers do churn each month. Customer churn bring loss in revenue for the company. ISP xyz want to know the prediction when the customer will do churn and the factors that affect customer churn. In this paper, customer churn analysis in Interne service provider company use data mining technique with
more » ... ral network method, because it produce result better than the others. Neural network combined with back propagation model and sigmoid biner function. Back propagation neural network's superiority is learning process that use weight rating point, if weight is not accurate it will do the learning process again. With that method, it will produce prediction customer churn result accurately. The result from this research is the most significantly effect factor for customer churn are the result of survey1, survey 2, and survey 3. Back propagation neural network has a accuracy 99.99% to predict customer churn. For reducing customer churn, strategy planning for CRM division is improving customer experience from the customer.
doi:10.15623/ijret.2016.0512027 fatcat:ehgkl3p5f5h6rbtj2sori2q3wm