摘要
在考虑了干散货运输需求、干散货船舶供给以及干散货船舶闲置等对运价指数影响的基础上,采用神经网络技术对月度干散货指数进行了研究,并且与基于ARCH模型和多元线性回归模型的预测进行了对比研究,结果表明从提高预测精度的角度来说神经网络技术是最优的.
This paper investigated the potential of neural networks for medium-term monthly dry bulk freight rates. A comparative study of predictive performance among neural networks, multivariate regression models and ARCH time series models were conducted. Our conclusions show that neural networks can significantly outperform multivariate regression models and ARCH time series models in forecasting performance.
出处
《大连海事大学学报》
CAS
CSCD
北大核心
2006年第2期67-70,共4页
Journal of Dalian Maritime University
基金
上海市高校科技发展基金资助项目(021K02)
