基于神经网络的股票预测系统研究

本文设计了一种基于粗集理论和神经网络的股票操作支持系统。系统根据对股票历史数据分析,预测股价未来一段时间内的走势,进而对投资者进行股票操作支持。指导投资者在投入资金一定的情况下,如何操作才会使总收益为最大。本系统首先利用粗集理论对预测数据进行属性约简等处理,然后把处理过的数据作为神经网络的输入。这样不仅减小了神经网络的规模,同时通过消除对象冗余减少了网络的训练和学习负担。与采用单技术的预测系统相比,本决策支持系统的可信度也有了较大的提高。

The Research of Stock Forecasting System which is Based on Neural Networks

The paper is a study to a stock operation support system which is based on neural networks and rough set theory. Accord-ing to the analysis to the history data of the stock,the system can forecast the stock's trend in future and guides the stockholders operate on the stock.It can also make the stockholders know how to operate to make the profit most under the condition that the as-sets is fixed. First, the system uses rough set theory to deal with the data to be forecasted with reduction of attributes.Second,it uses the disposed data as the inputs of neural networks.It reduces the scale of the neural networks as well as the training and studying load of neural networks with eliminating object redundancy.Compared with the systems which adopt the single technique,the system also makes the decision support confidence enhanced greatly.