基于ANN方法的股票预测模型

发布时间：2018-03-22 06:10

本文选题：股市预测　切入点：BP算法　出处：《重庆大学》2014年硕士论文　论文类型：学位论文

【摘要】：股票市场在当今经济活动中起着举足轻重的作用，是一个国家或地区经济和金融活动的“晴雨表”和“报警器”。研究股市走势对于国家和股民都具有举足轻重的意义。股市是一个多因素影响、高度非线性的系统，神经网络能够自动抽取数据集合中的非线性关系并进行模拟，因此常用于分析类似于股价预测的时间序列数据。本文对股票价格短期预测进行了理论与实证研究，建立了相应的BP神经网络，并针对其易陷入局部最优值、收敛速度慢、预测精度不高等缺陷，进一步使用具有全局搜索优化特点的遗传算法对神经网络的初始权值和阀值进行了优化，建立了GA-BP算法模型，并以长安股票进行了实证分析，并以MAPE误差评价指标对两种算法模型的预测精度进行了对比。实证结果表明GA-BP算法的预测精度较之BP网络有了明显的提高，但是隐层节点个数选择等问题仍需要进一步的理论研究。
[Abstract]:The stock market plays an important role in the current economic activities and is a "barometer" and a "warning device" for the economic and financial activities of a country or region, and it is of great significance to study the trend of the stock market for both the country and the investors. The stock market is a multi-factor, highly nonlinear system. The neural network can automatically extract and simulate the nonlinear relations in the data set. Therefore, it is often used to analyze time series data similar to stock price prediction. In this paper, theoretical and empirical research on short-term stock price forecasting is carried out, and a corresponding BP neural network is established, which is easy to fall into local optimal value and converges slowly. Because the prediction accuracy is not high, the genetic algorithm with global search optimization is used to optimize the initial weight and threshold value of neural network, and the GA-BP algorithm model is established, and the empirical analysis is carried out with Chang'an stock. The prediction accuracy of the two algorithm models is compared with the MAPE error evaluation index. The empirical results show that the prediction accuracy of the GA-BP algorithm is significantly improved than that of the BP neural network. However, the selection of hidden layer nodes still needs further theoretical research.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F830.91;F224

【参考文献】