摘要
针对都市圈货运量影响因素众多、样本量小、信息冗余的特点,用主成分分析法构造条件属性,用反双曲正弦函数处理货运量构造决策属性,进而将条件属性和决策属性作为新的空间样本输入量,构建改进型支持向量回归模型来进行训练和预测。然后,以重庆都市圈货运量为实例进行验证,并同多种经典预测模型进行对比实验分析。分析结果表明:用预处理后重新构造的空间样本作为输入量构建的支持向量回归机预测模型不仅能对都市圈货运量进行有效预测,而且预测精度很高。
For metropolitan freight many factors, the small sample size, information redundancy features, principal component analysis method is used to structure conditional attribute, the inverse hyperbolic sine function is used to process freight and structure decisional attribute. Thus the conditional attribute and decisional attribute as a new sample are input the amount of space to build an improved support vector regression model for training and forecasting~ Meanwhile, Chongqing metropolitan freight validated as an example is compared with the experimental analysis of a variety of classical forecasting models. The results show that after pretreatment reconstructed space as the input sample volume, the built support vector regression pre- diction model not only can effectively predict metropolitan freight, and its prediction accuracy is high.
出处
《世界科技研究与发展》
CSCD
2015年第4期390-394,421,共6页
World Sci-Tech R&D
基金
重庆市科学技术委员会决策咨询与管理创新重点项目(cstc2013jccx B0002)资助
关键词
反双曲正弦函数
货运量
预测
主成分分析
支持向量回归机
inverse hyperbolic sine function
cargo
forecasts
principal component analysis
support vector re^ression
