Applications of GRNN Based on Particle swarm algorithm Forecasting Stock Prices

摘要：

　　Generalized regression neural network (GRNN) has very good effect on making nonlinear forecasting model with large number of stock data.Particle swarm optimization (PSO) has simple operation analysis and is easy to implement.We use PSO algorithm to optimize the GRNN in order for optimal smoothing factor and connection weights.The prediction errors of the two models are both small.The MSE error by GRNN model reaches 0.0486,while the error by PSO-GRNN model is 0.0104.The analysis shows that PSO-GRNN model is more accurate,more stabilized and more generic than GRNN model.

关键词： generalized regression Particle Swarm Optimization neural network model Stock Price Prediction

作者: Jinna Lu Yanping Bai

作者单位: Jinna Lu,graduate students of NUC,research direction:math problems of the computer science,Mailbox 772,3 xueyuan RD,North University of China, Taiyuan, Shanxi Prov, China, 030051

会议类型: 国际会议

会议名称: 2013 International Conference on Information, Business and Education Technology (2013信息、商业与教育技术国际会议(ICIBET 2013))

会议地点: 北京

会议语种:英文

页码: 69-72

在线出版日期: 2013-03-14（万方平台首次上网日期，不代表论文的发表时间）