Data-Driven Air Quality Prediction with Batch Normalization in Long Short-Term Memory Networks

The health of every human is dependent on the level of oxygen which they inhale. Oxygen and its purity are dependent on trees around the living beings. Forests play an important role in the living beings as they increase the purity in oxygen. In urban, the deforestation leads to air pollutants which increases the health disturbances in humans. It is necessary for the human to know the cause of impurities in air and due to which toxins, the air gets impure at what rate. This type of research helps the industries, health organizations know and predict about the pollution in air and this contamination in air leads to which type of danger to human. The World Health Organization gets a support for predictions because of this research. Mainly air quality data is utilized for prediction purpose using deep learning approaches which is needed to make decisions and analysis of suitable data. The main aim is to provide users an effective prediction of air quality data on real time by univariate time series and a batch normalization in long short-term memory networks (BN-LSTM) which allows timely prediction of air quality using deep learning. The average magnitude of errors was improved to 28.06.