Spatial and Temporal Data Analysis with Deep Learning for Air Quality Prediction

Abstract

Air quality is an active topic at many social and political scales around the world. It is a significant concern for governments, environmentalists, and even data scientists who are raising awareness about this growing global problem. The availability of the massive amount of data in recent years enables better predictions of air quality using machine learning techniques. In this study, we perform spatial and temporal analysis using Long-Short Term Memory (LSTM) neural networks to estimate the nitrogen dioxide concentration that is considered a dangerous air pollutant between Beijing and London. In our proposed approach, spatial and temporal data are collected, preprocessed, normalised, and classified with LSTM followed by a comparative analysis with alternate machine learning techniques. The results show that the performance from our adapted approach of LSTM is higher compared to other techniques for predicting pollution rates between London and Beijing.