The Client Churn Analysis Based On Data Mining Techniques

Along with the open of telecom market, the contests for customers are getting more and more drastic. As the saturation of the telecom market, the cost of getting a new customer is much higher than maintaining an old customer. So, how to keep the customers, especially the valuable customers, came to one of the most important works for the telecom companies. As one of the important parts of the Analyzer Support System of China mobile, building the model of the churn users will allow the company to analyze the consume characters of those churned user, to find out those customers who are going to churn, so to take actions in time. So the study of this topic has very importance significance for reducing the cost of running the company and to improve the outstanding achievement of the company.In this thesis, according to the market condition that the Guizhou Communication Company is facing, using the data mining techniques, we create a client churn model which is based on the GuiZhou Mobile Analyser Supporting System. By using this model, we analysed the reasons why the clients churn, and also bring out the relevant client saving strategies. The main works including:1. A summarize of data-mining technologies clustering, correlation analyse for discoverable data mining and classification, data prediction for predictive data mining. Also describes the architecture of data mining, the process of data mining and the related data mining tools.2. Based on business analysis, the thesis brings out the principle for building a client churn model. By cleaning up and analyse the original data, build a churning model and churn prediction model on the Analyser Supporting system using data mining tools.3. Designed the index which representing the current client state, and data signals and related derived signals representing the possibility of client chum.4. Using the client churn model, the thesis have analyzed the character of churned user, and get the consume characters such as calling times, calling companies of a churned client, and grouped the clients according to above-mentioned characters, and also analyzed the main reason for clients to churn, and bring out the relevant client saving strategies.Along with the construction of Analyzer Support System and the development of data-mining techniques, the telecom companies will find more and more valuable client information and client consuming models, so they can have more efficient decision-makings.