基于相关性聚类对城市空气污染分析数据真实性的判别

Discrimination of Reality of Urban Air Pollution Based on Correlation Clustering Method

目的根据京津冀、长三角和珠三角等地区空气污染数据,构建数学模型分析数据的真实性,利用污染物之间的相关性、变化的连续性,判定各城市的空气质量数据是否存在不真实,并分析不真实的原因,为政府部门制定政策提供参考。方法以3个地区为研究对象,在互联网上查找到21个主要城市的空气质量污染数据,并定义不同污染物的污染指数API和空气质量分指数IAQI,使用相关性聚类的方法将污染程度相似的城市进行聚类划分。以AQI指数为因变量,以代表性污染物指标PM_(2.5)、SO_2和NO_2为自变量,进行多元线性回归,来判断观测数据的异常点。结果通过MATLAB和SPSS等软件对21个城市的空气污染数据处理分析,得出空气质量数据及污染物数据存在不真实情况。结论提出通过定期对空气质量数据进行观测排查来辨别分析数据的真实性,同时通过大力整顿工业园区来改善空气质量。

Objective For authenticity discrimination and analysis of air quality of urban air pollution data of 21 cities in Beijing-Tianjin-Hebei Region,Yangtze River delta and Pearl River delta,the authenticity of data was analyzed through modeling.Using the correlation between pollutants and the continuity of change,we can determine whether the city's air quality data is untrue,and make analysis of the reasons to provide reference for government departments to develop policies.Methods Taking three regions as the research object,the air quality pollution data of 21 major cities were found.On Internet,,and the pollution index API and air quality index of different pollutants were defined.Then was the correlation clustering method was used to cluster the cities with similar pollution levels.The AQIindex was the dependent variable,and the representative pollutant indexes PM_(2.5),SO_2,NO_2 were the independent variables for multiple linear regressions to determine observation data of abnormal points.Results The analysis of 21cities' air pollution data through MATLAB and SPSS software showed that the air quality data and pollutants data were untrue.Conclusion The suggestions were offered that we should regularly observe the air quality data to discriminate its authenticity,energetically overhaul industrial park so as to improve air quality.