Prediction of Delhi's Air Quality Index Using Deep Learning Approach

Air pollution is becoming a concern issue as it deteriorates the ambient atmosphere. On account of rapid industrialization and urbanization & wreaking havoc on the nature, this situation is getting worsen diurnally. Consequently, it causes health issues to human & animals, potential damage to historical monuments, and imparts negative impact on plant growth. Numerous initiatives have been made by state and federal governments to reduce air pollution. Globally, air quality is measured by the “Air Quality Index (AQI)” that further used to educate the public about the current status of air pollution in their surroundings. This suggested research explores an effective method for predicting Delhi’s air quality index. In this study, the 'Gated Recurrent Unit (GRU)' model is used to forecast daily air quality index. In this model, air quality index isused as a 0dependent variable & daily pollutant concentration as an independent variable. Furthermore, hyper-parameter settings have been also assigned such as No. of neurons - 50, epochs - 100, learning rate - 0.001, loss function –Mean square error (MSE), training & testing dataset – 80 – 20 (in %), andoptimization function–‘ADAM’. The accuracy of model represents through Mean square error (MSE)- 21.89, Root mean square error (RMSE)- 4.69, Mean absolute error (MAE)- 3.43, Correlation coefficient (R 2 value)- 0.99. Therefore, this study is proposing "GRU" model to produced precise &accurate daily estimates of air quality index