论文部分内容阅读
股票价格的预测是广大投资者非常关注的问题,也是诸多学者不断研究的方向,神经网络具有学习样本规律的特点,通过神经网络预测股票价格是近几年研究的重点之一。Copula EDA-BP混合优化算法是利用了copula EDA的全局寻优和BP算法局部求精的特点,将两者结合起来建立了基于copula EDA-BP的模型系统,优化神经网络的权值阈值,对股票上证180的收盘价进行预测得到误差率,结果显示copula EDA-BP算法平均误差率低于BP算法,提高了传统BP神经网络的计算精度。
Prediction of stock prices is a problem that investors are paying close attention to. It is also the direction that many scholars study continuously. Neural network has the characteristics of learning sample law. Prediction of stock prices by neural network is one of the focuses in recent years. Copula EDA-BP hybrid optimization algorithm utilizes the global optimization of copula EDA and the local refinement of BP algorithm. Combining the two, a copula EDA-BP-based model system is established to optimize the weight threshold of neural network. The result shows that the average error rate of copula EDA-BP algorithm is lower than BP algorithm, which improves the calculation precision of traditional BP neural network.