SDG-11.6.2 Indicator and Predictions of PM2.5 using LSTM Neural Network

Abstract

Smart cities can immensely benefit from the applications of Artificial Intelligence. These cities are highly attractive by their rich pull factors like the provision of facilities for safe and sustainable living. Sustainable Development Goals (SDGs) by the United Nations are the blueprint to improve the standards of sustainable living in all countries. The impact and achievement of SDGs are regularly assessed at country-level. To briefly describe a part of this process, we consider the current status of GCC countries regarding their achievements for SDG11.6.2 indicator that focuses on air quality. World Health organization regularly updates air quality database and when a source of reliable air quality data is missing, air quality in cities is modelled. We use LSTM neural network that learns from historical values of air quality data and predicts new values. This alternative approach may be used to confirm missing or inconsistent PM2.5 values. The objectives of our studies are to highlight one of the possible modern applications of AI to predict missing or unreported data and to leverage the concept of SDGs driven smart cities. We evaluate the performance of the LSTM model, and our results show that this model is capable of predicting data with acceptable accuracy.