Author : Shweta K. Borse and D. V. Patil

Air pollution is a serious problem. Pollution affects on human health and the atmosphere, it effects on health with diseases like cancer, asthma, heart disease and so on. An airborne pollutant can be described based on the absorption of elements available in the atmosphere. If the strength of a biochemical is larger than the goal level of elements in air, it is called as an air impurity. Airborne pollution arises when risky or extreme sizes of elements with smokes (such as CO, CO2, SO2, NOx, Ch4, PM) and Organic particles are familiarized into Earth’s air. Airborne pollution levels in utmost of the town areas has been a substance of thoughtful apprehension. In estimating of pollution, the soft computing methods are used. The air quality is predicted with machine learning algorithms that in-turn forecasts the AQI. AQI is a measure used to show the impurity levels over a time period. We have implemented the model to predict the AQI on previous year’s data of impurity. In this paper, we have proposed to use StackLSTM and as per our knowledge it performs better as compared to available techniques. We have used machine learning techniques such as a recurrent neural network (RNN), Long short-term memory (LSTM) i.e. SimpleLSTM and StackLSTM for experimentation. It is observed that StackLSTM performs better as compared to SimpleLSTM and simpleRNN.

Keywords: Air quality index, Simple Long short-term memory, StackLSTM, Simple Recurrent neural network.