Analysis of Air Quality Prediction Using Artificial Neural Networks

Air quality has utmost importance in understanding the status of Environment. The study below demonstrates effectiveness of Artificial Neural Networks (ANN) in environmental monitoring and prediction research. A multiclass dataset is taken from Kaagle. Tensorflow is the library used from which keras is imported to implement ANN. Various activation functions, optimization functions and dropouts has been used to obtain results. The highest obtained accuracy is 93.32% with least Loss of 0.18%. A graph is plotted to show the gradual increase in accuracy v/s decrease in loss using Matplotlib. ROC curve clearly shows the performance of each class in form of a graph. This paper highlights the effectiveness of Neural Networks in prediction of Air quality Index (AQI)