基于灰色关联的BP神经网络的沪深300指数预测研究

发布时间：2018-01-06 15:07

本文关键词：基于灰色关联的BP神经网络的沪深300指数预测研究　出处：《重庆师范大学》2013年硕士论文　论文类型：学位论文

【摘要】：由于股票的高回报以及高风险性，对于股票的研究和预测方法层出不穷。而股票市场是一非线性动态系统，传统的预测方法的预测误差较大，由于BP神经网络可以在有噪声环境下找到相对精确的结果且计算机的快速发展，使得依赖于计算机的BP神经网络在建立模型和预测方面得到广泛的应用。本文将使用BP神经网络预测沪深300指数，选取的是2012年8月1日到2012年12月31日之间的104天的数据，预测的输入量则选取股票中的技术指标，但由于技术指标众多，并不是每一个技术指标对股指的预测都有影响。所以本文采取灰色关联分析方法来选取预测指标，并且使用MATLAB7.1工具箱对BP神经网络预测模型的程序实现。设计三个方案，经过比较发现，BP神经网络的输入样本量不同，对于预测结果有一定影响；MATLAB中实现结果显示了经过筛选后的技术指标进行预测得到的结果更加接近实际值，并且输入样本太少误差较大，，但输入样本太多，年限跨度太大得到的预测精度也不会太高。
[Abstract]:Because of the high return and high risk of stock, the research and prediction methods of stock emerge endlessly. However, the stock market is a nonlinear dynamic system, and the prediction error of traditional forecasting method is large. Because BP neural network can find relatively accurate results in noisy environment and the computer develops rapidly, the computer based BP neural network has been widely used in modeling and prediction. In this paper, we use BP neural network to predict the CSI 300 index, and choose the 104-day data from August 1st 2012 to December 31st 2012. The predicted input is chosen as the technical index of the stock, but there are a lot of technical indexes. Not every technical index has an impact on the prediction of the stock index. So this paper adopts the grey correlation analysis method to select the prediction index. The program of BP neural network prediction model is realized by using MATLAB7.1 toolbox. Three schemes are designed. It is found that the input sample size of BP neural network is different after comparison. It has certain influence on the forecast result; The results in MATLAB show that the predicted results are closer to the actual values, and the input samples are too few, but the input samples are too many. The precision of prediction is not too high if the span of years is too long.
【学位授予单位】：重庆师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP183;N941.5;F832.51

【参考文献】