Font Size: a A A

Data Mining Applications In The Telecommunications Industry, Churn Analysis

Posted on:2008-04-14Degree:MasterType:Thesis
Country:ChinaCandidate:F WangFull Text:PDF
GTID:2208360212494045Subject:Operational Research and Cybernetics
Abstract/Summary:PDF Full Text Request
Along with Chinese telecommunication reform reorganization,The Chinese telecommunication industry has obtained the surmounting-like development, and has become one of the quickest development speed industries in the national economy, but behind development the deep level questions is becoming more and more obvious. According to the newest statistics of the information industries department, although the number of telecommunication customer increased recently two years in the telecommunication operation business of Chinese, the service income actually obviously has dropped. On the one hand the growth customer number is swift and violent, on the other hand the income earning of an enterprise is actually daily contraction. The situaiton that the development is more but loses money is more serious is worried. Whether is China's telecommunication experiencing the abundant harvestthe paradox? Looked from the surface that, "the increase does not additionally receive" is precisely the abundant harvest paradox characteristic, and its reason is that the telecommunication enterprise customer value is not high. This requested Chinese to transform the past traditional ideas, and should not only pay great attention to take the customer number finally is standard as the inspection achievement, meanwhile must pay great attention to this enterprise the customer value and the customer quality.The telecommunication market competition is intense day by day, and operation business becomes steady or the low speed growth from the high speed expansion. At the same time, the competition intensified also enable the telecommunication customer to have more choices. The new customer enters the net at the same time, also the large quantities of original customers drains from the net. The investigation data indicated the expense that develops a new customer is as 4 ~ 5 times as the expense that maintains an old customer cost. The 80 percent profit of the telecommunication enterprise comes from the 20 percent customers. Then, in front of unceasingly changes in the intense market and competition the market demand, how does the telecommunication enterprise reduce the customer draining rate in the greatest degree? One of commonly used methods is the data mining technology.The data mining technology is the most formidable data analysis method in the present data warehouse domain. Simply speaking, the data mining discovers the concealment rule through the establishment mathematical model method, then discoveres the rules that people did not know beforehand facing the magnanimous stored data,and then will be used in the practice. Data mining already had the successful application in very many professions.In this article data mining algorithm is used to resolve the drains of telecommunication enterprise. The tool used is Clementine software that is the more popular in the SPSS Corporation. The data warehouse provided by Shanghai telecommunication department is analysised. The algorithm used in the article is decision tree algorithm, Neuron network algorithm, clustering algorithm and association rule algorithm. Through establishing many mathematical models, comparing the precise of the models, and then we choose the best models—the model generated by C5.0, according to the result,I find that which kind of customers are easily drained,analysis the cause of the draining customers, and let the the corresponding telecommunication department take the corresponding measure in order to reduce the customer draining and enhances the economic efficiency. Making the reasonable wrap meal retents the customers that enter the net longer,and lets the older customers introduce more new customers. Understanding the demand of the customer, telecommunication enterprise can promote pack the service and overlapping sales model that meets the customer, and enhances the customer the loyalty and detains the customer; The goal customer category division is more clear, the telecommucition department enhances it's the quality of sevice , and these measures are more effective in the practice and decrease the drains of customers.
Keywords/Search Tags:data mining, decision tree, neuron network, clustering, association rule, stream
PDF Full Text Request
Related items