【摘要】 GMDH是一种具有自组织特征的数据处理方法,适用于非线性系统的建模,股指是一种重要的金融数据,具有混沌特性。该文将相空间重构引入了GMDH神经网络的建模中,并将之应用于道琼斯等股指的预测。同BP神经网络方法及一阶局域预测法相比,GMDH获得了更好的预测效果。更多还原

【Abstract】 The group method of data handling is a self-organizing data handling method,it is very apt to model non-linear system.The paper applies the revised GMDH method in prediction of the stock indexs which has chaotic character,comparing with the BackPropagation(BP) and the local prediction method of chaotic series with order one,it shows better results.更多还原