粗糙集属性约简方法在股票预测中的应用研究  

Research about application of rough set attribute reduction methods in stock prediction

作　　者：王天娥[1] 叶德谦[1] 季春兰[1] 

机构地区：[1]青岛理工大学中德信息技术合作研究所,山东青岛266033

出　　处：《计算机工程与应用》2009年第30期227-229,236,共4页Computer Engineering and Applications

基　　金：教育部留学回国科研基金No.0212498~~

摘　　要：针对神经网络在股票预测中遇到的困难,在预测模型中引入粗糙集理论,提出一种基于粗糙集与神经网络相结合的预测方法,并根据基本遗传算法的弱点对其进行了改进。首先,介绍了基于遗传算法的属性约简方法,对各遗传因子进行改进。然后,采用基于改进遗传算法的属性约简方法对模型的样本数据进行约简,删除冗余数据,得到样本输入的最小约简。最后,利用约简后的样本对预测模型进行训练与检验。实验结果表明,该方法具有较高的预测精度,能有效地解决网络结构复杂、学习速度缓慢等问题。Against the difficulties that the neural network encountered in the stock prediction,the rough set theory is introduced to the prediction model,and a new prediction method based on the combination of rough set and neural network is proposed.According to its weakness,the basic genetic algorithm is improved.Firstly,the attribute reduction method based on the genetic algo- rithm is introduced,and the genetic factors are improved.Then,adopt an attribute reduction method based on improved genetic al- gorithm to reduce the sample data of the model,delete the redundant data,and obtain the smallest reduction of the sample inputs.At last,the reduced samples are used to train and test the prediction model.The experimental result indicates that this method has high prediction accuracy,and can solve the problems of complex network structure and studying slowly and effectively. Key words： rough set theory ; attribute reduction; Radial Basis Function（RBF） neural network; genetic algorithm; stock prediction

关 键 词：粗糙集理论 属性约简 RBF神经网络 遗传算法 股票预测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]