An Empirical Analysis Of Shanghai Composite Index Based On Fuzzy Information Granulation And SVM Optimization Model

The coming of big data brings new challenge for data mining,how to mine useful information in a lot of data became a big project.Fuzzy information granulation model through indistinct information and make it particle as blurry granulation,to greatly decrease the complex of information,to simplify process of solution,to decrease computation.Membership function as fuzzy granulation limited rule mainly influences information granulation.This paper combines the fuzzy information granulation based on different membership functions with the support vector(regression)model to analyze the Shanghai composite index and to find the best membership function.The empirical analysis of the Shanghai composite index show that the hybrid models based on different membership functions have the same prediction results for the median R of the fuzzy particles.The accuracy and reliability of the hybrid model based on the asymmetric parabolic membership function is the highest.However,the prediction error of common prediction model which takes time t as an independent variable,is related to the fluctuation of the data.The prediction effect of the model is affected by the time nature of data,and the accuracy and reliability of the range prediction result can't meet the requirements.At the same time,the fluctuations of cycle coverage ratio and cycle reliable ratio between the two adjacent time windows are relatively large,and the stability of the model is poor.To eliminate the influence of the time nature of the data,this paper abandons the common hypothesis that time t is an independent variable,and introduces the daily highest point,daily minimum point,daily closing point,daily turnover(million hands)and daily turnover(billion)as independent variables,reforms the hybrid model.From the results of empirical analysis,this approach can effectively exclude the impact of the time nature of data on the forecast results,and significantly improve the accuracy and reliability of the range forecast.Meanwhile,the hybrid model has smaller fluctuations of cycle coverage ratio and cycle reliable ratio between the two adjacent time windows,the stability of the model is higher.