基于支持向量机与信息粒化的上证指数预测研究

Regression Prediction Method Based on the Chaotic Particle Swarm Optimization SVR and Information Granulation

参数的优化选择对支持向量机回归算法(SVR的预测精度和泛化能力影响显著,提出混沌粒子群优化算法(CPSO优化选取支持向量回归算法中参数c和g信息粒化是进行海量数据挖掘和模糊信息处理的有效工具。在此基础上利用上证指数数据建立上证指数开盘数预测模型,研究结果表明,混沌粒子群优化的SVR信息粒化时序回归预测模型克服了传统时间序列模型仅局限于线性系统的缺点,速度快,预测精度高,且实用性强。

Optimization of the parameters significantly affect prediction accuracy and generalization abilityof support vector machine regression （SVR） algorithm.This paper proposed chaotic particle swarmoptimization algorithm （CPSO） the optimal selection of support vector regression algorithm parameters cand g, and information granulation is effective tools for massive data mining and fuzzy informationprocessing. Use the data of shanghai Composite Index to establish the opening number of predictivemodels. The research results show that chaotic particle swarm optimization SVR granulation time seriesregression prediction model to overcome the traditional time-series model is limited to the shortcomingsof the linear system, have a speed high prediction accuracy and practicality.