A CSI 300 Index Prediction Model Based on PSO-SVR-GRNN Hybrid Method

In this article, a PSO-SVR-GRNN nonparametric hybrid model is proposed for the CSI 300 stock index to forecast problem. Particle Swarm Optimization (PSO) utilized optimize parameters of SVR enhance prediction ability support vector machine's regression original Index time series. The optimized residual sequence results General Regression Neural Network (GRNN) are then used series prediction. outcomes indicate that PSO- SVR-GRNN can greatly improve accuracy compared with individual models such as PSO-SVR, GRNN, GA-SVR, LSTM, PSO-LSTM, and SVR.