Methods Of Extension Data Mining And Their Applications

With rapid development of computer technologies and increasing applications of database management systems, the accumulated data in modem life are far beyond our normal capabilities in analyzing and understanding them without the use of automated analysis technologies. In fact, important information may be found behind the accumulated data. Therefore, it is necessary to systematically and deeply analyze these data, which provide clues for scientific decisions. Data mining has been developed to solve the above challenges. Data mining, an information extraction process, can be used to explore hidden facts out of the database. Out of question, data mining has become one of the most heated topics in the field of information technology.With the advancement of global economics, the renewal period for the information and knowledge has been shortened because of the rapid environmental change. Innovations and solutions of problems become more and more important in various fields. Therefore, how to mine extension knowledge becomes critical for the research of data mining. The research indicated that extension data mining shows promising applications in many fields. Extension data mining is a combination of extenics and data mining. It is used to explore the knowledge about extension transformations in databases, i.e., extension knowledge which takes advantage of extension methods and data mining technologies such as extension classification knowledge, extension clustering prediction knowledge, extension association rules knowledge, and conductive knowledge.After reviewing and studying the available approaches on data mining reported, this thesis discovers an effective and creative model, i.e., Extension Data Mining model, with a combination of extenics, fuzzy theory, rough set theory, sets pair analysis, and other data mining technology. The main results of the thesis are summarized as below:(1) Research on extension clustering predictionBased on the complexity of traditional forecasting methods, an extension clustering prediction model has been built up by combing extenics and clustering method. Firstly, attribute reduction was achieved for clustering with similar attributes through the hierarchical clustering method; secondly, the extension clustering for the remaining attributes were made by the use of their changing rates, and weight coefficient was determined with a simple correlation function; finally, the results show that it is available to predict indicators of China Unicom by using extension clustering prediction method. Therefore, this method may be beneficial for making decisions and expanding markets.(2) Research on customer propagation values based on extension methodsFaced with the shortcomings of current customers’ evaluation and thedeficiencies of combined quantitative and qualitative descriptions, this thesis has subdivided customers’ values with the use of conjugation analysis combined with extension and customer value theory, and the thesis also provides customer propagation values. The customer propagation value model has been built up through both qualitative and quantitative methods. The results present beneficial references to make marketing strategies and promotional images.(3) Research on extension data mining in subdivisions of telecommunication enterprise brands.Faced with the complexity and subjectivity of the current attribution reduction as well as the determination of weight coefficient, sub-brand model has been created with combination of extenics and rough set methods. There are some attributes which are unimportant to the decision attribute, and some records that disturb on making decisions. Reducing the condition attributes based on the matter-element theory and rough sets pair analysis, the thesis has calculated the importance of the decision attribute for each condition attribute after reduction. The thesis has also determined weight coefficient through the use of rough set methods and relevant experience. This work may be beneficial on how to integrate the existing brands and how to recommend an appropriate brand to new customers. Through the analysis of three indicators of brands of China Unicom, the results show that extension data mining can provide effective support for the Decision-making of enterprise and appropriate service differentiation.(4) Research on extension association rulesIn connection with static characteristics of current association rules and combination with extension methods and association rules, we have firstly analyzed the positive extension field, the negative extension field, the positive stable field, the negative stable field, and the extension boundary owing to extension transformation. And then we analyze the change association rules and various situations of the changes of conditions and conclusions, and_present support and confidence. The result shows it is available to analyze China Unicorn’s package by using extension association rules. Research results provide beneficial references to marketing strategies.Finally, a brief summary and some future research directions are highlighted in this dissertation. Research results provide beneficial references to decision-making, customer relationship management, marketing strategy and promotional image. This thesis is very important and beneficial to both theoretical research and engineering practice.