Predictive modeling and analysis of air quality - Visualizing before and during COVID-19 scenarios.

Quality air to breathe is the basic necessity for an individual and in recent times, emission from various sources caused by human activities has resulted in substantial degradation in the air quality. This work focuses to study the inadvertent effect of COVID-19 lockdown on air pollution. Pollutants' concentration before- and during- COVID-19 lockdown is captured to understand the variation in air quality. Firstly, multi-pollutant profiling using hierarchical cluster analysis of pollutants' concentration is performed that highlights the differences in the cluster compositions between before- and during-lockdown time periods. Results show that the particulate matter (PM and PM) in air that formed the primary cluster before lock-down, came down to close similarity with other clusters during lockdown. Secondly, predicting air quality index (AQI) based on the forecasts of pollutants' concentration is performed using neural networks, support vector machine, decision tree, random forest, and boosting algorithms. The best-fitted models representing AQI is identified separately for before- and during-lockdown time periods based on its predictive power. While deterministic method reactively evaluates present AQI when current pollutants' concentration at a particular time and place are known, this study uses the best fitted data-driven model to determine future AQIs based on the forecasts of pollutant's concentration accurately (overall RMSE<0.1 for before lockdown scenario and <0.3 for during lockdown scenario). The study contributes to visualize the variation in pollutants' concentrations between the two scenarios. The results show that the reduced economic activities during lockdown period had led to the drop in concentration of PM and PM by 27% and 50% on an average. The findings of this study have practical and societal implications and serve as a reference mechanism for policymakers and governing bodies to revise their actions plans for regulating individual air pollutants in the atmospheric air.