基于遗传算法神经网络集成股票市场预测研究

神经网络集成技术能有效地提高神经网络的学习能力和泛化能力,已经成为机器学习和神经计算领域的一个研究热点.本文利用不同的神经网络算法产生神经网络集成个体,以误差平方和最小为准则,用遗传算法动态求解集成个体的非负权重系数,进行最优组合集成建模研究,并以此建立股市预测模型.通过上证指数开盘价、收盘价进行实例分析,计算结果表明该方法相对传统的简单平均集成模型,具有预测精度高、稳定性好,易于操作的特点.

Study on the Stock Market Prediction Model of Neural Ensemble Based on Genetic Algorithms

Neural Network ensemble can significantly improve the learning and the generalization ability.Recently,neural network becomes a hot topic in machine learning and neural computing application.In this paper,many different neural networks are first generated by different training algorithms. Secondly,the non-negative weighted of each Neural Network ensemble individual is thus obtained using Genetic Algorithm dynamic solving.This method is established for the forecast model of Shanghai Stock Exchange index.Finally,the example is applied to the forecast model of Stock Market.The experimental results show that the proposed approach can effectively improve the prediction accuracy and stability.