Prediction of air-pollutant concentrations using hybrid model of regression and genetic algorithm

Abstract

Air pollution is one of the major environmental concerns in recent time. The majority of the population in the developed world live in urban area, hence air pollution concern is even more in cities. The worst gaseous pollutants are Caron Monoxide (CO), Nitrogen dioxide (NO2) and OZONE (O3). In this paper, we propose two predictive models for estimation of concentration of gases in the air, namely Carbon Monoxide (CO), Nitrogen dioxide (NO2) and OZONE (O3). The first proposed model is a combination of linear regression and Genetic Algorithm (GA). The second proposed model estimates concentration of gasses using Multivariable Polynomial Regression. First model uses a linear regression for prediction of concentration of gases, whereby errors like MAPE, R2 obtained by linear regression are optimized using a genetic algorithm (GA). Multivariable Polynomial Regression is adopted as a second proposed method for the prediction of concentration of same gases. A detailed comparative study has been carried out on the performances of GA and Multivariable Polynomial Regression. In addition, predictive equations are formed for CO, O3 and NO2 based on temperature, relative humidity, benzene and Nox (oxides of nitrogen).