Research On Securities Investment Based On Data Mining Technology

With the rapid economic development in China, chowmatistic investment is becoming more and more important for people. Moreover, stock index futures and margin loan& stock loan enrich the means of investing. Under the circumstance, stock market certainly will attract more individuals and institutional investors for interest holders benefit, reduction of hedging risk and value maintenance. In consideration of that, no matter in practical application or theoretical research, the study of analysis method of stock investment must affect the market positively. However the traditional analysis method is powerless in the complex market, which has massive and day-by-day growing data, and changes constantly.Based on the theory of data mining and its related theories, this paper would give intensive study in analysis method of stock investment. Initially, wavelet theory would be introduces to the procedure of financial data preprocessing. Considering wavelet could particularly processing non-stationary and nonlinear signal, it should wipe off the noise in the financial data, and it also have more positive results and affects on data mining in the future. Then, as to regard the yield of listed companies, this paper set up enterprise lifecycle distinguishing model, which is based on Navie Bayes classier and principal component analysis theory and combined with financial index, for estimating which step of the lifecycle is the listed company on. After this, this paper discusses about merit and demerit of Neural Network comprehensive evaluation model, and improves RBFNN with factor analysis theory. Meanwhile, F-RBF neural network comprehensive evaluation model is set up for assessing and calculating yield of stock. Finally, this paper applies wavelet analysis, enterprise lifecycle distinguishing model and F-RBF neural network comprehensive evaluation model to the stock market in China to analyse 50 listed companies practically.The innovation points of this paper are as followed:1.Wavelet would be applied in. finance and stock data denoising, and, comparing to the traditional denoising methods, it would have more positive results and affects on data mining in the future. 2. The defect of Navie Bayes classier would be conqured by principal component analysis. Moreover, modified Navie Bayes classier would be set up, which integrating with selected financial index, builds up enterprise lifecycle distinguishing model.3. The original defect of RBF neutral network would be overcome by factor analysis theory. Meanwhile, F-RBF neutral network comprehensive assessment model is set up to assess stock yield of listed companies in China.