摘要
论文研究基于神经网络的股票预测方法,针对目前存在的问题,通过模糊理论与动态神经网络的结合提出一种更为适合现状的动态模糊神经网络DFNN(Dynamic Fuzzy Neural Network)股票预测模型。首先对采集的股票信息进行属性提取,然后利用粗糙集理论中的信息熵算法进行属性约简、删减冗余信息,最后用约简后的数据作为动态模糊神经网络的输入属性进行训练预测,并在算法模型中运用分级学习的思想,能在一定程度上实现预测某一只股票短期内大致走势的功能。实际操作中更能为股票的多重选择进行推荐,降低投资的风险,有着较高的实用性。
This paper studies the stock prediction method based on neural network and proposes the dynamic fuzzy neural network(DFNN)stock prediction model which is more suitable for the current situation through the combination of fuzzy set theory and dynamic neural network. Firstly,feature extraction is performed on the collected information,then the information entropy is used to simplify the attributes and remove the redundant information in the rough set theory. Finally,the obtained attributes are used as the input of the dynamic fuzzy neural network to perform training predictions,and to the certain extent,using the idea of grading learning in the algorithm model which can achieve function of predicting the general trend of a stock in the short term. In practice,it can recommend multiple options for stocks,reduce the risk of investment and have a higher practicality.
作者
王峥
温光洒
邱秀连
WANG Zheng;WEN Guangsa;QIU Xiulian(Nanjing FiberHome StarrySky Communications Development Co.,Ltd.,Nanjing 210019;Wuhan Research Institute of Posts and Telecommunications,Wuhan 430074;Nanjing FiberHome TianDi Communications Technology Co.,Ltd.,Nanjing 210019)
出处
《计算机与数字工程》
2020年第3期517-522,共6页
Computer & Digital Engineering
基金
国家重点研发计划“智能服务交易运行机理及性能优化技术”(编号:2017YFB1400704)资助。
