Air quality index prediction via multi-task machine learning technique: spatial analysis for human capital and intensive air quality monitoring stations

Air, an essential natural resource, has been compromised in terms of quality by economic activities. Air pollution has become a critical environmental issue in recent decades. Forecasts of air quality play an important role in warning people about and controlling air pollution. Considerable research has been devoted to predicting instances of poor air quality, but most studies are limited by insufficient longitudinal data, making it difficult to account for seasonal and other factors. Several prediction models have been developed using an 8-year dataset collected by Beijing Environmental Protection Monitoring Center (BEPMC). The small spatial and temporal scales and nonlinearity of climate effects are significant challenges to precise Air Quality Index (AQI) prediction. As a result, data normalization is applied to enhance air quality features without distorting divergences. Machine learning methods, including simple linear regression (SLR), support vector regressor (SVR), random forest (RF), and probabilistic voting ensemble, were used to build regression models for predicting the AQI of six central and outskirt regions of Beijing city. The mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R-2) were used to evaluate the performance of the regression models. Experimental results showed that the max probabilistic voting ensemble performed better in the prediction of the AQI in terms of R-2, whereas RF performed better in the prediction of the AQI in terms of MAE and RMSE scores. The overall performance of the proposed model in terms of MAE and RMSE is between 0.0128 to 0.0194 and 0.0230 to 0.0326, respectively. This work also illustrates that combining ensemble learning consisting of various classifiers' output weights with air quality prediction is an efficient and convenient way to solve certain significant environmental problems.