Predicting PM2.5, PM10, SO2, NO2, NO and CO Air Pollutant Values with Linear Regression in R Language

Air pollution is one of the most challenging and complex problems of our time. This research presents the prediction of air pollutant values based on using an R program with linear regression. The research sample consists of obtained values of air pollutants such as sulphur dioxide (SO2), particulate matter (PM10, PM2.5), carbon monoxide (CO), nitrite oxides (NO, NO2, and NOX), atmospheric data pressure (p), temperature (T), and relative humidity (rh). The research data were collected from the city of Belgrade air quality monitoring reports, published by the Environmental Protection Agency of the Republic of Serbia. The report data were transformed into a form suitable for processing by the R program and used to derive prediction functions based on linear regression upon pairs of air pollutants. In this paper, we describe the R program that was created to enable the correlation of air pollutants with linear regression, which results in functions that are used for the prediction of pollutant values. The correlation of pollutants is presented graphically with diagrams created within the R GUI environment. The predicted data were categorized according to air pollution standard ranges. It has been shown that the derived functions from linear regression enable predictions that are well correlated with the data obtained by automatic acquisition from air quality monitoring stations. The R program was created by using R language statements without any additional packages, and, therefore, it is suitable for multiple uses in a diversity of application domains with minor adjustments to appropriate data sets.

Automated summary

This research uses an R program with linear regression to predict air pollutant values. The data used in this study were collected from air quality monitoring reports in Belgrade, published by the Environmental Protection Agency of the Republic of Serbia. The results show that the derived functions from linear regression can effectively predict air pollutant values.