Hushen 300 index forecasting approach based on ISNN and HGA
基于ISNN和HGA的沪深300指数预测方法*

This paper proposed an improved structure-based neural network (ISNN)and applied to construct a forecasting model for Hushen 300 index. Designed an outstanding hybrid genetic algorithm (HGA)and used to train the ISNN forecasting model. Evaluated the proposed approach by the Hushen 300 index of the first half year at 2007. Experimental results suggest that the proposed approach has more favorable characteristics such as the convergence rate, learning ability, forecasting precision and estimating error.