Using Data Mining Techniques To Establish A Credit Rating Scale And Application Of Mobile Phone Users

The Taiwan telecommunication has been liberalized over 20 years. Today Taiwan mobile phone operators are limited by all sorts of factors. It is difficult to have the large scale growth again in getting new subscribers. The maintaining cost of one subscriber is 1/5 of getting a new subscriber. Therefore, how to manage subscriber loss will be the most important task for every mobile phone operator.According to the subjective wish of the subscribers, it can be divided to the voluntary churn and forced churn in the telecommunications industry. The forced churn is also called the passive churn. It means that unpaid bill or broken the contract. To terminate the services is proposed by the telecommunication enterprise, not by the subscribers. The subject of the thesis will focus on the forced churn.Since it is forced churn, it can be the "risk subscriber" for the operators. Therefore, it is a big challenge for the operators to manage the forced churn loss, rise subscriber's satisfaction, reduce the bad debt, and maximize the business revenue. Is the subscriber who has the high contribution but has been forced churn a good subscriber? On the other hand, is it not worth to give service or to maintain for the low contribution's subscribers? What kind of subscribers should be regard as beyond questions or an alarm?... All questions above involve if the subscribers should be scored and classified.This research content takes a Taiwan mobile phone operator as an example. How does the mobile phone operator use the data warehousing technology to establish and process the credit score model under the situation of facing the saturation of the market, getting the subscriber difficultly, decreasing ARPU, slowing revenue growth. Finally, the " risk level" has been set up by the "credit score" and average ARPU of the contribution of each subscriber. The "risk level" can be used in adjusting the internal work flow, the cost and the service level; and further, it can help the company to assign the limited resources effectively.After two year experiences, the thesis thought that the "credit score model" not only can have the alarm functions, avoid number of amount of bad debts and revenue loss. It also can have a remarkable effect upon "increasing the telecommunication's revenue" and "rising the customer satisfaction".