基于神经网络的股票预测【摘要】: 股票分析和预测是一个复杂的讨论领域，本论文将股票技术分析理论与人工神经网络相结合，针对股票市场这一非线性系统，运用 BP 神经网络，讨论基于历史数据分析的股票预测模型，同时，对单只股票短期收盘价格的预测进行深化的理论分析和实证讨论。本文探讨了 BP 神经网络的模型与结构、BP 算法的学习规则、权值和阈值等，构建了基于 BP 神经网络的股票短期预测模型，讨论了神经网络的模式、泛化能力等问题。并且，利用搭建起的 BP 神经网络预测模型，采纳多输入单输出、单隐含层的系统，用前五天的价格来预测第六天的价格。对于网络的训练，选用学习率可变的动量 BP 算法，同时，对网络结构进行了隐含层节点的优化，多次尝试，确定最为合理、可行的隐含层节点数，从而有效地解决了神经网络隐含层节点的选取问题。【abstract] Stock analysis and forecasting is a complex field of study. The paper will make research on stock prediction model based on the analysis of historical data, using BP neural network and technical analysis theory. At the same time, making in-depth theoretical analysis and empirical studies on the short-term closing price forecasts of single stock. Secondly, making research on the model and structure of BP neural network, learning rules, weights of BP algorithm and so on, building a stock short-term forecasting model based on the BP neural network, related with the model of neural network and the ability of generalization. Moreover, using system of multiple-input single-output and single hidden layer, to forecast the sixth day price by BP neural network forecasting model structured. The network of training is chosen BP algorithm of traingdx, while making optimization on the node numbers of the hidden layer by several attempts. Thereby resolve effectively the problem of it.【关键词】BP 神经网络 股票预测分析1．引言股票市场是一个不稳定的非线性动态变化的复杂系统，股价的变动受众多因素的影响。影响股价的因素可简单地分...
