多元线性回归分析建立空气质量数据校准发布的数学模型

Mathematical model for air quality data calibration and release with multiple linear regression analysis

本文通过对国家监测控制站点和某公司空气质量检测仪数据的研究,运用Excel数据分析工具箱进行多元线性回归分析,找到国控点和自建点的数据关系。运用MATLAB多元线性回归函数,建立国控点和自建点的数据模型,利用国控点数据对自建点的11项监测数据进行校准发布,并对所建模型进行共线性判断、残差分析、异常值诊断,对样本数据的初筛和在数据拟合时如何剔除异常值与不显著线性项提出可行性方法。

Through the research on the data of the national monitoring and control station and the air quality detector of a company,this paper uses Excel data analysis toolbox to carry out multiple linear regression analysis,and finds the data relationship between the national control point and the self built point.By using the multiple linear regression function of MATLAB,the relationship model between the national control point and the self built point data is established.The 11 monitoring data of the self built point are calibrated and released by using the national control point data,and the collinearity judgment,residual analysis and outlier diagnosis are carried out for the built model.The feasibility method is put forward for the initial screening of sample data and how to eliminate the outliers and the non significant linear item in the data fitting.