| With the economic development and the improvement of people’s income, thestock market has been called more and more people’s attention，Stock investment hasbecome an indispensable part of people’s lives. So far, randomly selected stocks havebeen unlikely to be rewarded, and along with the promotion of the efficient markethypothesis theory, a passive investment strategy is gradually being recognized by themajority of investors. But there is no denying that there are always some stocks whichthe value is being undervalued in the stock market. Therefore, in sufficient competitivemarket, investors want to get benefits greater than the average return on the market,therefore the study of stock prediction method has greater application value andtheoretical significance. At the same time, the stock market always accumulated a largenumber of data information passed through many years development. How to use thedata information in the investment decision-making has become the focus of the stockmarket prediction.Support vector machine (SVM) is a new technology in data mining. In this paper,the SVM classifier was applied to stock selection. First, this paper introduces thestatistical learning theory and the basic principles of SVM are discussed.Second, in order to improve the accuracy of stock selection, financial fraudidentification SVM model is constructed in this paper, so as to discovery exception ofthe listed company’s financial position and warn investors, thereby protecting theinterests of investors. Moreover, in this model, which breaking the traditional1:1matching method, using principal component analysis method control sample, andSVM was introduced for unbalanced data processing method to obtain a highaccuracy..Finally, in this paper, the deep Shanghai stock100was chosen, the result issatisfactory. Because of the (SVM) method has the advantages of itself, with thedeepening of the research, it will become an important way to solve the problem ofstock market prediction. |
PDF Full Text Request