武汉市PM2.5影响因素多元回归分析.doc

摘要
本文对武汉市2013 年1 月—2013 年8 质量浓度和影响因素数据资料进行整理统计,对影响大气污染的各个污染指标进行综合分析, 质量浓度与其他污染指标存在不同的相关性,这类模型具有一定的实际应用价值,分
别采用“强行进入法”“逐步进入法” 指标的多元线性回归模型,比较了
对缺失值进行不同的方法处理时,差异不是特别大;还有就是共线性的问题,多重共
线性并没有影响到整个模型的拟合,因而不会对模型得到因变量的点估计值有影响。
通过定性分析,定量计算以及对各物理量之间的相互作用过程研究, 质
量浓度变化特征和各影响因素之间的关系。结果表明,对于5 个基本指标,两两之
间,出了臭氧与二氧化氮以及臭氧与可吸入颗粒物, 的水平上都
是显著相关的;通过F 的水平上, 与其余指标均是显著
相关的,其中, 与臭氧呈现负相关,与其余量呈现正相关,与
一氧化碳的相关
系数最高;在最后进行了检验分析。在进行验证时,我们利用线性模拟,二次模拟分
别与原始数据进行比较,得到的拟合效果比较好,是我们所要的结果;在最后面,根
据线性多元回归模型得到的结果, 的
影响是最大的,提出了一些相应的措施,能够有效地控制大气污染。关键词: 污染特征相关性回归分析
万方数据
华中科技大学硕士学位论文
II
Abstract
In this paper, Wuhan January 2013 2013 August mass concentrations and
factors influencing collate statistical data , effects of atmospheric pollution on various
pollution prehensive analysis of mass concentrations were established
with other pollution indicators have different correlations such models have some practical
value respectively, using "enter" " stepwise" to establish a multiple
linear regression model
pared to the missing values different approach , the difference is not
particularly large ; there is collinearity problem of multicollinearity does not affect the fit
of the entire model , and thus will not get the model variables because the point estimate
values affected. Through qualitative analysis , quantitative calculation as well as the
physical interaction between the various studies , the relationship between mass
concentration derived characteristics and the influencing factors. The results show that for
the five basic indicators between any two out of ozone and nitrogen dioxide , and ozone and
particulate matter , the rest of the index at the level were significantly associated ;
through F hypothesis test results in on the level , and other indicators were
significantly correlated , in whi