PREDICTION OF AQI USING HYBRID APPROACH IN MACHINE LEARNING

Abstract

Forecast of urban air pollutant concentrations must deal with a growth in the volume of environmental monitoring data as well as complex changes in the air pollutants. This initiates the need of efficient prediction techniques to increase prediction accuracy and stop pollution causing issues. In this paper, the combination of Random Forest Regression (RFR) and Support Vector Regression (SVR) based machine learning model is proposed to predict the AQI values. Daily meteorological and air pollution data between March 2020 and June 2022 from Jind city in Haryana were used for model training and test the model’s results. First, the important factors affecting air quality are extracted and null values are replaced by their mean values to handle the irregularities in the dataset. Second, the Random Forest regression machine is used for model training and prediction of the value, the SVR model is used in correction of residual items, and finally the predicted values of AQI are obtained. The experimental results showed that the proposed prediction model of RFR-SVR had a better prediction result than the standard Random Forest Regression, support vector regression machine learning. RMSE, R2, MAPE are used as evaluation indicators to evaluate the performance of the proposed model.