Application And Research On Data Mining In Customer Relationship Management

Presently, applying Data Mining technology to Customer Relationship Management is the key to advance the decision-making efficiency of enterprises, which may extract useful information from large numbers of customer data, help enterprises predict customers' action trend and support management strategy. And also by using this technology enterprises could unfold activities to meet the demand of their customers and provide them with the right products and services. The thesis roundly analyzes and researches the theory and knowledge of Data Mining and Customer Relationship Management, construes the basic principle and characteristic of Data Mining algorithms in many fields, and demonstrates the idiographic application of Data Mining algorithms in Customer Relationship Management.The optimization of decision tree classification algorithm is the most important research content of this thesis. After some disadvantage of the classic decision tree classification algorithm of ID3 is analyzed, a new algorithm of ODT-BID3, which can optimally build a decision tree, is given. By selecting the maximal average value of information gain as the standard of testing attribute, the algorithm gets over the question that the ID3 algorithm tends to choose an attribute having more value. Besides by combining some repeated branches of a decision tree in the cause of building it, the algorithm conquers another question that there exists many reduplicate data in a decision tree built by ID3. After the theoretical correctness of the ODT-BID3 algorithm is proved, the thesis analyzes its performance, time and calculation complexity degree in mang aspects. Finally, the instance testing is implemented to examine the 0DT-BID3 and to contrast it with the algorithm of MID3 which is also an improved algorithm on the base of ID3. The testing results indicate that the decision tree optimally built by the algorithm of ODT-BID3 is distinctly better than the tree produced by MID3. And the algorithm not only can build a decision tree with the optimized structure and rather scale, but also can help enterprises mine out better rules information and sequentially offer decision-making support for them. Furthermore, through establishing a classification model, the thesis researches and discusses how to apply the algorithm of ODT-BID3 to Customer Relationship Management.