Research On Stock Market Forecasting Model Based On RS-LS-SVM

Support vector machine(SVM)as a common data mining method,it has a very prominent advantages compared with other methods,the current application in various fields is also very wide.However,there are many problems in the practical application of the study worthy of study,the model itself has a lot of parts can be improved.In order to further improve the support vector machine model and promote it,this paper chooses the capital market which has many research on the using of support vector machine but still have some problems as the research object.Based on the analysis of the existing stock market forecasting methods and the shortcomings of the standard support vector machine,this paper proposes a least squares support vector machine model based on rough set to forecast the stock market.Firstly,we use the rough set to attribute the forecasting index and then use the least squares support vector machine to predict the fluctuation of the stock price,in order to provide some reference for the stock market forecast,and provide the idea and method of the application for the rough set and the least squares support vector machine.The main contents of this paper are as follows: Firstly,the present research situations of stock market forecasting,support vector machine,least squares support vector machine and rough setare introduced systematically.And on the basis of previous studies,this paper sums up the shortcomings of the existing methods.In view of the shortcomings of the existing methods,the RS-LS-LSVM stock market prediction method has been put forward;Secondly,the theory foundations of rough set and least squares support vector machine(SVM)and the selection of kernel function have been discussed;Then,according to the whole idea of the model,the flow chart of the model is established,and the process of the forecasting model is described in detail.And set up a set of the forecast index system which include 27 indicators,such as the latest models including today's highest price,yesterday's highest price,the highest price of the day before,and the average price of 7 days,et al.,and the various steps in the model of the process and methods were described one by one;Finally,randomly selected China Petroleum(601857),Huilong shares(002556)and Jin Sheng Precision(300083)2016 annual trading data,in the main board,small plates and the GEM,each forecast is 244 groups Data,in the MATLAB software for three times the three groups of models of the comparative experiment,each comparison experiment has carried out 20 random tests,respectively,in the software using RS-LS-SVM,LS-SVM and RS-SVM on the sample data Regression prediction,and the results were compared.Experimental results show:First,the experiment for three samples obtained three different reduction index system,indicating that the same indicators for different predictors,the effective indicators are different;the same indicators on the application of different predictors are also different.Therefore,it is necessary to carry out index screening and attribute reduction before forecasting.Attribute reduction can reduce data redundancy and improve forecast performance,and the primary index system proposed in this paper can predict China's stock market to a certain extent.Second,the experimental results verify the feasibility and effectiveness of the RS-LS-SVM prediction model.Whether it is in the motherboard,small plates or in the gem,the number of experimental results show that the performance of MSE and RMSE two data,RS-LS-SVM prediction model is better than LS-SVM and RS-SVM model.We can see that the rough set and the least squares support vector machine are introduced into the stock market,which simplifies the difficulty of the model and improves the speed of the solution.It has some innovation,which has the characteristics of the stock market forecasting model and the investors' reference value.