Loss Analysis, Business Intelligence-based Telecommunications Customers

With the development of telecom competition, there are more and moreopportunities for customers to select the telecom products. Snatching customersbetween telecom enterprises is drastic more and more. Based this state, most of theoperation companies are realizing the importance and the necessity of customer missinganalyzing to development of enterprises.This article based on Sichuan Telecom. The author has surveyed the telecom fieldand researched and developed the telecom project almost one year and accumulatedmuch experience. The author studied many intelligence theories like data warehouse,OLAP, data mining and some relevant theories about KPI. By summarizing the defaultsof the former construction methods of customer missing, the author gives the new ideaof customer missing analyzing model which based business intelligence. This methodovercomes the subjectivity of the former methods. In the former methods, the peopleconstruct model used the key attributions supplied by business people subjectively. Butin the new method, the attributions of construction are analyzed by author inimpersonality. The article concludes computing customer missing degree value to findKPI of customer missing. First, give customer missing an accurate business definitionand finding all the possible attributions that influence the customer missing. Then thecustomer missing analyzing subject data warehouse was build. And then using OLAPtechnique deletes some unimportant attributions. Fourth, Constructing BP neural net andusing Sensitivity analysis and 80/20 principle analyze the key attributions. ConstructingBP neural net again making the key attributions as input and the customer missingprobability as output, training the BP neural net using training data sets until it isconverged .Until now, the customer missing analyzing model is successful.At end of the article, the author validated and evaluated the model using the reallydata sets from Sichuan telecom. The result proves the idea of entire intelligence solutionis very good. The result will effect the customer missing analyzing of Sichuan Telecom.