The Research Of Data Mining Classification Arithmetic In CRM

The development of market economy, the progess of science and technology, together with the informationalization drive the corporation into an unprecedented era of competition. Enterprises are only communicating with customs through various kinds of channels so as to gain knowledge, know more about customers and improve their satisfaction to obtain more profits, could win in the fierce market competition. Enterprise's competitive strategy has already changed from product-centered to customer-centered and the customer has already become enterprise's essential successful factor and profit source. Customer Relationship Management (CRM) can help enterprises understand the customer's demand better, improve their relation, improve their satisfaction, thus promote the competition strength of enterprises.This dissertation is started from the theory of CRM and Data Mining Classification. And then it dissertates the meaning, the basic content and the system structure of the application system of CRM, Data Mining Classification's course and data pretreatment. The Data Mining Classification is described especially in this dissertation, and apply it in CRM customer's classifying, customer's loss predicting.When dissertate the Data Mining Classification, the dissertation has studied the SLIQ classification algorithm. SLIQ classification algorithm through the technologies such as pre-sorting and scope-priority can deal with a large number of data set fast and accurately, and has better scalability and parallel. The dissertation detailedly introduces the attribute split method, pruning technology of SLIQ classification, and provides SLIQ classification's total procedure.The customer classification is the important component of analytic type CRM. The "marketing one to one" that sets up on customer classification is that enterprises improve the customer's satisfaction and increase the earning capacity essential means. To the telecommunication enterprise's characteristic thedissertation use SLIQ classification algorithm sets up customer's classification indexs and carries on customer's classification demonstration research.The neural network has some limitations such as converges slowly, apts to fall into some partial-purpose optimization, so introduce Genetic Algorithm (GA) to optimize the neural network. Using GA's better general-purpose optimization ability and suitable for parallel processing to optimize the neural network, and then structure the Genetic Algorithm Neural Network (GANN). In order to achieve the purpose that to optimize both the weight and the topological structure of BP neural network at the same time, the dissertation has give a GANN algorithm that base on three floors of chromosome structure, and apply it in China Mobile to predicting the customer's loss. Using the Data Mining Classification to predicting CRM customer's loss can effectively reduse enterprise losses that because customer's loss and can improve the competitiveness of enterprises greatly.