Application of Linear Neural Network Model to New Shanghai Composite Index Predication
Kui Wang · Journal of Chongqing Technology and Business University · 2012
Linear neural network model is used to make fitting predication for New Shanghai Composite Index(000017).By selecting total 66 data of the monthly data of New Shanghai Composite Index from January,2006 to June,2011,by using former 62 data as training group and latter 4 data as predication group,through comparing error square sum of the model in different time-lag windows,proper window numbers are selected as optimal model,the initial data of the model are optimized in order to promote simulation effect and then predication analysis is conducted.Results show that fitting effect is very good,except that the vibration of stock market is bigger in June,the monthly fitting error of other months is less than 3 percent.Short-term predication feasibility for stock market is elaborated.