Telecom Industry Customer Segmentation Based On Telephone Records

With the rapid development of modern information-based society,the important information is possibly hidden within the companies' informational database.We may analyze the mass data with the help of Big Data technology to find useful information,which can provide references to make strategies of enterprise.The telephone records of customers in the telecom industry contains a lot of unintuitive information,using big data technology to analyze customers' historical data can mine the characteristics and internal relations of customers' consumption behaviors,which will help with customer segmentation,so as to form personalized services for different kind of customers that has important practical significance and application prospect.This thesis is based on a part of an ongoing research project of China unicom limited Shaanxi weinan branch,and the data is provided by Shaanxi weinan unicom.The main work of this thesis can be summarized as followings:1.A database is designed with Oracle,which is used to collect basic call information.2.Selecting Clementine software to collect the customer call history data of the three major operators of China mobile,China telecom and China unicom(Shaanxi weinan unicom data mainly,others' data for comparison),and processing data to extract customers' characteristics.3.Using three kinds of clustering algorithms to establish models that can classify customer categories and summarize their characteristics,and the model of the optimal algorithm was selected and evaluated.4.After clustering,C4.5 decision tree algorithm was used to divide customers into three categories of high value,medium value and low value accurately,so as to provide targeted personalized services.5.Using the concept of fuzzy set correlation degree to analyze the correlation degree of two calling Numbers,and realizing the visual interface of correlation degree query,the customer service staff of communication operators can directly query the relationship between customers through simple operation.6.After customer segmentation,designing a system of customers' personalized service,which is embedded with customer classification subsystem and relational degree query subsystem.Operators can use simple button operation to complete the corresponding query.Through the customer's telephone record data analysis,we can see that the designed operating system can help operators for customer segmentation accurately and summary the behavior characteristics of the same customers,which will help operators to divide different services into different customer groups,so as to meet the needs and habits of customers to the maximum extent that will provide important references for personalized services.At the same time,through the correlation query subsystem,it can help to further expand services by the link between the customers.