A Systematic Review and Comparative Study of Machine Learning Techniques for Air Quality Prediction

Due to rapid urbanization and industrialization, air quality monitoring and prediction in metropolitan cities have become critical for improving public health. In contemporary research, machine learning models are being widely used for air quality index prediction as they can efficiently handle the complex relationships existing in the vast amount of pollutant data. This paper presents a systematic review of nine machine learning and deep learning algorithms focusing on prediction of Air Quality Index. The algorithms are compared on the basis of the environmental dataset of the city Bengaluru, India collected from the website of the Central Pollution Control Board of India (https://cpcb.nic.in/). The empirical study indicates that traditional machine learning algorithms cannot effectively capture complex temporal and non-linear relationships in datasets, while deep learning algorithms like Bi-directional LSTM with 1-dimensional convolutional neural networks significantly outperform others due to their ability to capture intricate temporal and non-linear relationships. The study offers a valuable insight by highlighting the advantages and drawbacks of the algorithms, and thus contributes to the existing knowledge in the field of air quality prediction using machine learning algorithms. Future work will explore advanced deep learning architectures, such as attention mechanisms and ensemble algorithms to further enhance AQI prediction accuracy.