The Analysis Of Statistical Models In Data Mining

With the research on data mining further deepened, people have come to realize that the research on data mining based on DMKD has three technical supports, namely, database, artificial intelligence and statistical algorithm. Statistics, as a technique, cannot be ignored in the field of data mining. Although the probability models based on the thoughts of statistical inference developed in the 20 th are complex in structure, they enable us to get more functional and more easily explained results.Based on many Chinese and foreign articles, this thesis mainly elaborates on audit data, tectonic model and other steps in data mining to offer reasonable analysis methods based on statistics. Firstly, the thesis focuses on the classification, induction and summarization of the statistical models. From many perspectives, it makes a classification of the statistical models based on the probability and mathematical statistics and also provides their applications. Secondly, it deals with the evaluation criteria of the statistical models and theoretically analyzes the characteristics and application scope of several models to offer specific methods about how to choose proper models. In addition, from multivariate statistical analysis together with principal component and factor analysis, the thesis provides an effective and easily explained data sieving method and its practical applications. Finally, it is the summary of the whole thesis and the future prospect of the application of the statistical analysis methods in data mining is also discussed.