Recognition Of Mobile Family Customers Based On Mobile Communication Interaction Circle

Domestic telecommunication industry has commenced full-service competition.China Mobile faced increasing pressures in mobile voice services. Family customers andgroup customers will be the new business and the major revenue growth points. How toidentify family customer in massive customer base, and analysis family customers businessbehavior characteristic. It’s an urgent topic. Without considering the establishment of amining model, based only on the physical discriminant rules of family customers andbusiness customers, we can extract only1.7million families around the customer,accounting for about7%of total customers. It’s far from the target of family customerbrand and number of products. Therefore, the demands of family customer mining whichusing the application of data mining techniques is imminent.The thesis solved the problem that prediction accuracy rate is not high and does notcover all issues of user groups with simple mathematical statistics. Based on unified datawarehouse of Shanghai Mobile and induction characteristics of users’ communication linkby establishing MX interaction circle, It’s different from the ordinary intercourse circlemodel, which using voice interaction circle superimposed SMS interaction circle andcombined with interaction index formula. It added weight calculation when combininginteraction circle model. According to known user information of interaction circle as input,it established the method of customer characteristics recognition. The thesis summarizedand classified different behavioral characteristics of mobile users with decision treealgorithm, and identified family customer’s consumption characteristics, behavior rules and the relationships between various factors. It established family customer identificationmodel by these relationship and filtered out potential family customers in communicationusers, through exchanges on valuation and consolidation methods to sort final extractdomestic customers. Finally, it extracted the full flow system of the internet searchkeywords of family customers by E-IDF method. It calculate the degree of matchingkeyword between keyword vector, and provided family customers subdivision standardswhich meet the needs of mobile communications companies, and accurately grasp theactual content to the customer preferences, value-added services to carry out precisionmarketing. Entire model helps to improve marketing and customer stickiness hit rate andloyalty, will also bring new revenue growth.Through experimental results, compared with the original user judgment based onlyon the basic properties, family customers screened by this model have greater degree ofimprovement in accuracy through the results of experimental tests. Running from ShanghaiMobile officially environment situation, the recognition model is able to identify andanalyze the existing operating system combined with running the family business classcustomers full control, The model can be combined with the existing traffic analysissystem through formal environment running situation. It also can monitor the full-servicebehavior of family class of customers, and take use of the customer tags to precisemarketing. From the management benefit point of view, due to timely constructed familycustomers recognition model, domestic market is different from rural market as it did notcarry out such a comprehensive diagnostic manual work, and saving a lot of manpower andresources to successfully promote a family customers meticulous management. The modelimproved meticulous management of home users, reduced monitoring waste of resources,and improved the efficiency of enterprise management. From the results of the social point of view, the model can achieve precise marketing and won the respect of customers byavoiding duplication selecting the target user group. It achieved scientific development offamily communication market.