Short-term and Long-term Air Quality Forecasting Technique Using Stacked LSTM

For the entire globe, air pollution has been a worrying issue. Earth's Atmosphere contains numerous toxic gases and harmful solid particles are caused by Air pollution. Contaminated Air have been many mischievous effects on human health. Asthma, emphysema, chronic obstructive pulmonary disease and lung cancer can happen due to air contamination. Among other enlisted polluted cities, Dhaka lies in a hazardous problem for air pollution. This paper has approached two Long Short-Term Memory (Vanilla LSTM, Stacked LSTM) model and Gated Recurrent Unit (GRU) model to Predict air Quality Indexing with different hyper-parameter tuning. And analyze future the health effects based on Air Quality Index Level. We have worked on Dhaka Air Quality data which was collected by the United States Environmental Protection Agency (EPA). Among the two models, we have acquired the highest accuracy of 91.61% for Short-term prediction and 90.83% for Long-term prediction. And RMSE value of 4.65 and 16.19 for Air Quality Index value prediction on Stacked LSTM tuned with 200 hidden nodes on the first layer and 100 nodes on the second layer.