Data Mining Research Based On The Extension Of The Clustering Method

At present, the ability that we use advanced technology such as computer internet is greatly improved that it used to be. Much data and knowledge were applied in commerce strategic decision, marketing analysis, science study, and project development and so on, and this trend will get sustainable development. As the modern society information technology fast develops and the scale and scope of data application expands unceasingly, the data that people get is becoming more and more and the kinds of the data also become increasingly. Especially the rapid development of the Internet brought a lot of data and information for we, it becomes a urgent need to solve important problem that how to get implied, useful information or knowledge for commercial decision from such large-scale and containing abnormal data database information and further improve the information utilization. In this case, it becomes extremely important to study data mining and its methods. This thesis is based on this point; it proposes a new method called the Extension of the Clustering Method after analysising and studying the methods and algorithm of the data mining.This thesis is studying on knowledge about the extension project and data mining. It is based on the related research achievements of predecessor both at home and abroad, analysising DM basic theory and cluster methods, introducing the extension theory and knowledge into data mining, start from basic thought, tools and method, making the problem formal, and found extension cluster method. This method is based on the matter element, defining the knowledge as the matter element. First, converting the knowledge into the form of matter-element model, then forming extension set, and definite classical domain and joint domain, finally found dependent function, and using the size of value of dependent function to judge the degree that the knowledge information belong to the classification, and finishing the cluster. The works this thesis studies are including the following aspects:(1) It has more detailly discussed about general state of the data mining related theory and its application research both at home and abroad, including the general situation, characteristics, process, several kinds of DM methods, and the applications of DM in science research, finance, medical treatment and so on.(2) Doing further research on cluster analysis method of data mining, including the general state of cluster analysis, the content, advantage, disadvantage of five kinds of common cluster analysis, describing several kinds of main cluster arithmetic and comparing them from the time complexity,the attributes of target data, the shape found by cluster, the sensitivity of abnormal data the sensitivity of data input sequence, the high sensitivity and the efficiency of the algorithm.(3) Discussing about the theories, definitions and formulas of extenics which are used in setting up extension cluster methods, including basic element theory, extension set theory and dependent function; then raising the general process during building extension cluster method model and elaborating to describe its content; finally based on the detail data, verifying the validity of extension cluster method model by using earthquake classification distinguishing instance.