A Novel Air Quality Prediction System with Long Term Storage with Hyper Parameter Tuning

Air pollution is the most important contributor to a variety of serious health problems as well as weather transformation. Air quality interpreters are needed to design human activity at a particular environmental location and lessen the adverse pollution detrimental impacts. The authors have examined the problem in difficulty of forecasting the classification of pollutant concentrations in the future and have proposed a novel method based on Long Short-Term Memory (LSTM) model. This is a Deep Neural Network (DNN) model that is identified to work well with consecutive prediction difficulties. Through the data obtained, this approach produces a prediction model that reliably forecasts the Air Quality Index (AQI) from Central Pollution Control Board (CPCB) website. LS TM is experimented eight number of times to choose the best possible function using Python’s LeakyRelu package to assess hyperparameter optimizations. The comparative analysis of accuracy metrics such as R Squared, Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) are measured for different models. The proposed model discovered R Squared is more than 1, indicating that the hyper parameter tuned model is the best fit model, based on the concurrent experiments. This dataset has the highest prediction accuracy and less prediction loss.