Prediction of air quality levels to support sustainable development goal 11 using multiple deep learning classifiers

Purpose – Sustainable Development Goal (SDG) 11 emphasizes the importance of monitoring air quality to develop cities that are resilient, safe and sustainable on a global scale. Particulate matter pollutants such as PM2.5 and PM10 have a detrimental impact on both human health and the environment. Traditional methods for assessing air quality often face challenges related to scalability and accuracy. This paper aims to introduce an automated system designed to predict air quality levels (AQLs). These levels are categorized as good, moderate, unhealthy and hazardous, based on the air quality index. Design/methodology/approach – This paper uses a dataset of 8.1 million air quality records from various US cities. The data undergoes preprocessing to remove inconsistencies and ensure uniformity. Scaling techniques are applied to standardize the values across the dataset. Augmentation methods, including K Nearest Neighbour, z-score normalization and Synthetic Minority Oversampling Technique (SMOTE), are employed to balance and enhance the dataset. Later, the data are used to train eight deep learning models, including standard, bidirectional and stacked architectures. Additionally, two hybrid models are also developed by combining features of different architectures. Findings – The validation results demonstrate the system's exceptional performance. The Bidirectional GRU model achieves the highest accuracy of 99.98%. Similarly, the hybrid model RNN + Bidirectional GRU achieves an impressive accuracy of 99.92%. Furthermore, the Stacked Gated Recurrent Unit stands out, achieving perfect scores of 100% for precision, recall and F1 score. Originality/value – Traditional air quality assessment approaches rely heavily on basic statistical methods and are limited by the scope of their datasets. In contrast, this study presents an innovative methodology that employs advanced deep learning models and hybrid architectures. By incorporating sophisticated preprocessing techniques, the proposed system significantly enhances the detection and classification of AQLs, setting a new benchmark for achieving sustainable development objectives.