Identification of Safe Air Quality for Activity Using Long Short-Term Memory

Abstract

Air pollution is one of the problems that is often occurred in the cities with high industrial and transportation levels. Several informations are needed by the community so that they can be use it for people’s health concerns. Air Quality Index is a standard for determining air quality and pollution based on the main parameters, namely: Carbon Dioxide (CO2), Sulfur Dioxide (SO2), Ozone (O3), and Particulate Dust (PM10 and PM2.5). While, for the methods, we suggest Long Short-Term Memory model which have good performance in identifying safe air quality for activities. The model conducts training with a parameter of 100 epoch, learning rate = 0.001, batch size = 32. Testing uses the optimizer, activation function, and the right function to measure air quality with an accuracy of up to 98%. Thus, we adjust LSTM model with Adamax optimizer and Sigmoid activation function is the best parameter to get the highest accuracy.