Some New Methods Of Data Mining And Their Applications In Stock Market Of Our Country

Data mining(DM) is the technology which develops fastest in information fields. Many experts such as statistician,database expert all acquire the developing space, so data mining increasingly become a live topic in business circles. With the development of information technology, the way by which people collect data increasingly rich and brilliant, and the quantity of data increasingly expand. The data level even reach to GB or TB, and high dimension of data has become the mainstream. These plentiful data and their high dimensional character make the traditional data analysis method be outshone. With the capability of computer increasingly renew, people expect that computer help us analyse and understand data, and help us make the right decision base on abundant data. Mathematical statistics is one of the most important and active subjects in applied mathematics field. It come into being before the computer be invented, and have developed for hundreds of years. Now strong and valid statistical methods and devices have become the base of information consultation. In information age, consultant trade becomes more prosperous. But the combination between mathematical statistics and database technology is't fast, the aggregate function is so simple in database query language SQL, and this is a theory. It is far from enough that consultant trade inquire data by database. Theory of probability and mathematical statistics has new vigor once people have a request that from data query to knowledge discovery,from data deduction to data induction, so a scene of prosperity present on the crunode of DM. Stateside SAS company is famous for mathematical statistics method and visualization calculate proclaim that it place him in the middle of DM, and it illustrate this point. So it is necessary to apply statistics knowledge to DM field, and let DM get abundant development, and let applied cost of statistics get abundant embodiment. This paper explore some statistical methods, and apply them to Shanghai and Shenzhen stock-market. The first chapter mainly introduce some related notion of DM,background and state of the art at home and abroad. In the second chapter, we apply some mature method to do farther analysis in Shanghai and Shenzhen stock-market based on Yan Jinan[1], and give a strong judgment that home stock-market is't efficacious, and provide basic precondition for following business. The third chapter apply correlation theories of linear model to bring up new method to detect outliers, and acquires better result by empirical analysis. The fourth chapter apply some distance method(Cook distance method,likelihood distance method) in outlier diagnostic, and give a new Cook distance based on this. We mine efficient "linear"and "unlinear"ranges. In last chapter, we apply curvature method to mine valuable "unlinear"ranges, especially the mine effect of quadratic form unlinear ranges is very good.