安徽工程大学毕业设计（论文）
中国货运量预测方法研究
摘要
货运量是确定物流需求的一个重要指标，是为确定物流基础设施建设规模和制定各项相关政策决策的一个主要依据。货运量预测结果的合理性和可靠性将直接影响到物流基础设施的投资收益和相关物流企业的发展，并对地区资源合理的分配以及物流发展战略的制定都具用非常重要的意义。
本文分为六个部分，第一部分是绪论部分，阐述了本论文的研究背景、选题的意义、研究内容和研究方法，并分析了国内外的研究现状，使读者了解了撰写本文的意义及相关研究的发展水平;第二部分，概述了时间序列平滑预测，在这里我们应用了三种平滑预测方法，分别对中国历年的货运量做了频繁或预测;第三部分，采用了多元线性回归的方法，我们从统计年鉴中查找了影响货运量的几个指标的数据，这里我们共有四个自变量和一个因变量，并采用逐步回归的方法，找出影响货运量最显著地变量。第四部分，对中国货运量序列采用了ARIMA模型进行拟合，最终找到了适合该序列的模型为ARIMA（1，1，0）。第五部分，基于上面几种预测方法之后，对本文的最后一章，采用了灰色模型预测。从以上几种预测方法中得到模型预测最精确的是灰色预测模型预测值2013年为544415.04为第六部分是结论和展望，对本文结论进行了总结，并对后续研究中需要进一步解决的问题进行了展望。
本文主要在查找国内外知名学者的研究基础上对中国2012年以前的货运量数据做了详尽的研究，本文虽然在预测方法上并没有突破前人的方法理论，没有一味的追求算法的复杂性和先进性。但能够从不同的预测背景出发，同时这也是本文的创新点，即根据各种预测算法的适用范围及优缺点，合理的选择预测方法，这样才能够保证预测结果的可靠性。对每一种方法所对应的预测做出最终的预测结果。
关键词:货运量；预测；灰色预测；ARIMA模型；多元线性回归；时间序列平滑预测
I
张创：中国货运量预测方法研究
Abstract
Cargo is an important indictor for determining the logistic needs of the region. It is also the main basis for determining the scale of the logistic infrastructure and making industrial policies.The reasonable and reliable results of the Cargo Forecast will have a direct impact on the inventincome of region's logistic infrastructure and the development of related logistics enterprises. It is
also of significant meanings in the distribution of resources in the region and making strategies forthe logistics development.
There are six parts in this thesis. The first part introduces the backgrounds, aim, significance, main content and methods used in the research, and literatures related. Then, The second part, an overview of the time series smoothing, here we use three kinds of forecast methods of smoothing, respectively on the freight volume of Chinese calendar made frequent or prediction. , The third part, using the method of multiple linear regression, we find the impact of several indexes of freight from the statistical yearbook, here we have four variables and one dependent variable, and by the stepwise regression method, find out the most significant variables influencing freight volume. The fourth part, the Chinese freight traffic volume time series using ARIMA model, finally found a suitable for the sequence model for ARIMA (1, 1, 0) .The fifth part, after the above product prediction method based on, to the last chapter of this paper, using the grey prediction. Through comparing the results, the author finds the combination forecast model is a reliable model. Finally, the author draws a conclusion and shows the further research.
The basis of this study mainly for domestic and foreign well-known scholars on China before 2012 freight volume data to do a detailed study, although this article theory and method in the prediction method does not break the previous,not blindly pursue algorithm complexity and advanced. But from the point of forecasting background different, according to the scope of application of all kinds of algorithms and the advantages and disadvantages, prediction method of rational choice, so that it can ensure the reliability of prediction results. Prediction corresponding to each method to make the final prediction results.
Keywords: Cargo;Forecast,;grey forecasting; ARIMA model; multivariate linear regression; time series smoothing
II
安徽工程大学毕业设计（论文）
目录
摘要 ............................................................................................................................................. I ABSTRACT .............................................................................................................................. II 第1章 绪论 ......................................................................................................................... - 2 -
1.1研究背景和意义 .............................................................................................................................. - 2 - 1.2国内外研究现状 .............................................................................................................................. - 2 - 1.3数据来源 .......................................................................................................................................... - 4 - 1.4 研究内容和方法 .............................................................................................................................. - 5 -
第2章时间序列平滑预测法 ............................................................................................... - 6 -
2.1移动平均法 ...................................................................................................................................... - 6 - 2.2指数平滑法 ...................................................................................................................................... - 7 - 第3章多元线性回归模型................................................................................................................... - 10 - 3.1 多元线性回归模型理论介绍 ........................................................................................................ - 10 - 3.3 实证分析 ........................................................................................................................................ - 12 -
第4章时间序列ARIMA模型的建立和预测 .................................................................. - 17 -
4.1平稳时间序列分析ARIMA模型 ................................................................................................. - 17 - 4.2差分运算 ........................................................................................................................................ - 17 - 4.3 ARIMA模型 .................................................................................................................................. - 18 - 4.4实证分析 ........................................................................................................................................ - 25 -
第5章 灰色G（1，1）模型的建立和预测 ................................................................... - 29 -
5.1 灰色系统和模型的介绍 ................................................................................................................ - 29 - 5.2 灰色模型的建模步骤 .................................................................................................................... - 29 - 5.3 实际数据代入及参数运算 ............................................................................................................ - 31 - 5.4灰色模型预测 ................................................................................................................................ - 32 -
结论与展望 ......................................................................................................................... - 33 - 致 谢 ................................................................................................................................... - 34 - 参考文献 ............................................................................................................................. - 35 - 附 录 ................................................................................................................................. - 37 -
附录A 一篇引用的外文文献及其译文 ............................................................................................. - 37 - 附录B 列出主要参考文献的题录及摘要（10篇） ........................................................................ - 48 - 附录C 主要源程序 ............................................................................................................................. - 52 -
III
张创：中国货运量预测方法研究
插图清单
图2-1原始货运量趋势图........... ...........................................................................................8 图4-1原始时间序列的时序图.........................................................................................25 图4-2一阶差分时序图............. ...........................................................................................26 图4-3自相关图.....................................................................................................................26 图4-4白噪声检验........................ ............................................................................. ..........26 图4-5偏自相关图................ ................................................................................................27 图4-6模型的参数估计及残差白噪声检验.. ......................................................................27 图4-7模型的最终的预测............. .......................................................................................27 图4-8模型的最终的预测.....................................................................................................28
IV
安徽工程大学毕业设计（论文）
插表清单
表1-1原始货运量数据........................ ....................................................................................4 表2-1各种方法预测结果........................... .............................................................................9 表3-1各种指标的原始数据........ ................ ............ .............................................................12 表3-2 逐步回归方程................... ...........................................................................................13 表3-3 拟合优度检验.......................... ............... ...................................................................14 表3-4 方程的显著性检验........................... ..........................................................................14 表3-5 回归系数检验表...................... ...................................................................................15 表3-6 残差序列自相关性检验.................................... .........................................................15 表5-1 2003到2009的时间响应函数预测值.................... ………………………...............32 表5-2 模型精度检验............................ .................................................................................32 表5-3灰色G（1，1）预测........................ ..........................................................................32
V