Air Quality Index Prediction Using DL

Abstract

Abstract: Air pollution emerged as greatest environmental threat to human health and the planet. Occurred due to release of harmful substances like PM2.5, PM10, oxides of nitrogen, carbon and sulphur, ozone, volatile organic compounds etc.. into air by various human activities like vehicle exhaust, agricultural practices, cooking, combustion, burning of fossil fuels, use of air conditioners, refrigerators and other sources adversely effecting climate, ecosystems, health and biodiversity. Assessing and monitoring air pollution levels became obligatory. The current study deals with air pollution data collected from various cities in India over a period of six years (2015-2020). Deep learning models LSTM and GRU are employed to predict the air quality index. The dataset underwent meticulous preprocessing, followed by a sophisticated feature selection process utilizing Principal Component Analysis (PCA). This strategic approach allowed us to discern and isolate the principal pollutants exerting a substantial influence on air quality. The application of LSTM and GRU models for AQI prediction holds great potential in improving our understanding of air pollution dynamics.