Applied Research Database Mining In Telecommunication Customer Analysis

Integration and rapid development of global economic integration, computer scienceand technology science, the Internet and in related fields, the cloud platform scaleinformation sharing and discovery, has changed the traditional sense on theimplementation of enterprise development strategy model and marketing tactics. With thesustainable and healthy development of enterprises, particularly those based on theamount each household sector, a large amount of information in the database, deleteselection and digging for information, has become an important research focus incomputer software science staff. Therefore, this paper based on intelligent data miningtechniques based on the establishment of cross-selling mutual correlation algorithms,decision trees customer response, three types of telecommunications users clusteringalgorithm analysis model, and the corresponding theoretical and practical research projects.The main work done as follows:(1) The basic concept for database mining depth analysis, combining data miningprocess and telecom industry applications and common techniques of data mining, datamining technology and development trends of current research directions are summarized.(2) The establish of telecommunication customer cross-sell model to analyze theintrinsic relationship telecommunications value-added services. According to thecharacteristics business the corresponding affairs of the database, After analysis the choiceof FP-growth algorithms. Taking into account the large amount of data telecommunicationsfeatures, construct a memory-based FP-tree is unrealistic proposed based on PL/SQL’sFP-Growth algorithm, namely through the FP-growth algorithm into a relational databasesystem, the use of database management system itself powerful query capabilities, Searchfrom storage FP-tree database table sets out all the frequent patterns, reduce datatransmission and conversion time and the amount of data to solve problems. Then takeadvantage of the confidence, the judgments of association rules, the effective value-addedtelecom services to identify cross sell combined results to guide decision-makers takeeffective strategies..(3) The establish of a client-based decision tree response model. C provincialtelecommunications company has a variety of business, each business opened to extract themain features of larger customer base, which in turn can recommend the service to theprospective customers with similar characteristics. Corresponding decision tree modelbased on established customers, is to analyze the opening of "Music Raider" Most of thenatural attributes of user characteristics and consumer behavior characteristics, and thenbased on these characteristics, targeted marketing, which can greatly reduce the cost of the company to broaden the business market. Yiju SPRINT Suanfatedian and Youshi,Nenggoushixian Jue Ceshu one two Cha Xiaoguo, Tigao Shu Juku Juece Shu Jianlishijianand Xiangguan Fenleisudu, C Sheng Dianxingongsi Shuju Ku Shuojuwajue Jishu SPRINTSuanfayonghu Xiangyingmoxing, Jingguo Fanfushiyan, Queding the Heli’s threshold,extract database data classification principles to achieve C provincial telecommunicationscompanies use the "Music Raider" user features. SPRINT algorithm model and comparedwith the C4.5algorithm, the model evaluation results show that, SPRINT algorithm C4.5algorithm shows those of its exact value is slightly smaller, but has a relatively databasedata was statistically significantly higher coverage. In the Under the Business BackgroundC the provincial telecommunications company information’s, For customers predictionissue is concerned response, pair subscribe to the business clients’s recognition ability ofrequire a higher, that is to say model is pair coverage rate of ratio on precise the rate ofthere are higher requirements. So say relatively speaking from the prediction effect,SPRINT algorithm achieve better prediction effect.(4) The establishment of a database clustering algorithm based on user segmentationmodel. Demand for commercial purposes refers to the diverse needs of the user-oriented,with a commercial value of the target database data mining, this paper for telecomenterprise database business objective is to establish corporate users related subdivisions.First, a list of design for data analysis, data processing and database data by formatconversion; secondly, the application of K-Means algorithm to optimize the database usersegmentation model. Finally, the actual program to communicate through the actualsimulation and C provincial telecommunications business, and finally, to determine thevalue of the database data mining subdivision category, defined as four, and obviously hasa good consumer segmentation feature, database data type and class of significantdifferences between the establishment of the database data mining clustering algorithmmodel is feasible, database segmentation model-based clustering algorithm, has arelatively stable targeted, telecommunications products and services in favor of expandingthe market, a market core competitive advantage.