摘要
研究表明,大多时间序列可分解为线性部分和非线性部分的组合。融合ARMA模型在捕捉线性关系的优势和神经网络模型强大的非线性映射能力。介绍一种ARMANN集成模型,并对上证指数月收益率进行预测。结果表明集成模型的预测能力显著优于传统单一模型,对指数收益率预测比较准确,有很强的应用价值。
It has been proved that most time series can be broken up into a combination of a linear part and a non-linear one.This paper introduces an integrated model integrating the unique strength of two models in linear and nonlinear modeling and applies it to forecast the return rate of shanghai stock index.The result shows that the inte-grated model makes better prediction of the return rate of index than that of the single model and it has value of ap-plication.
出处
《兰州石化职业技术学院学报》
2014年第3期32-35,共4页
Journal of Lanzhou Petrochemical Polytechnic
