基于大数据背景下的多层神经网络股票预测模型  

Multilayer Neural Network Stock Prediction Model Based on the Big Data

作　　者：丁美琳 高语越 陈学斌 

机构地区：[1]华北理工大学,唐山063210 [2]唐山市数据科学重点实验室,唐山063210

出　　处：《软件》2017年第7期118-121,共4页Software

摘　　要：随着互联网以及股票市场的不断发展,产生了蕴含丰富信息的海量股票数据。由于大数据技术不断普及,处理海量股票数据逐渐变得可能。本文通过对海量的历史数据进行研究,利用智能算法建立多层神经网络对数据进行处理。首先运用小波分析技术将股票价格波动曲线分解为低频部分和高频部分,其次分别利用Elman和BP神经网络进行训练,最后进行小波重构得出股票价格预测值。研究结果表明:通过改进,将预测结果与实际值进行对比,均方误差MSE=6.4495′10^(-6),模型预测精度较好。With the continuous development of the internet and the stock market, there is a huge amount of stock information which contains abundant information. With the growing popularity of big data technology, dealing with massive stock data is becoming possible. In this paper, through the research of the large historical data, the data processing is established by using the intelligent algorithm of multi-layer neural network. Firstly, the stock price fluctuation curve is decomposed into the low frequency part and the high frequency part by the wavelet analysis technology, followed by Elman and BP neural network training. Finally, wavelet reconstruction is used to obtain the stock price forecast value. The results show that the prediction results are compared with the actual values by means of improvement. The mean square error is MSE=6.4495×10-6, and the prediction model has better accuracy.

关 键 词：股票价格预测 小波分解与重构 BP神经网絡 Elman神经网絡 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]