AI-Driven Air Quality Index Prediction

Abstract

The rapid increase in urban growth, industries and the rising number of vehicles has made it a major issue for many cities where air pollution is becoming a serious concern. It impacts people’s health but also damages the environment.” All existing systems show the current air quality, but do not predict pollution levels. Due to this, there is a requirement for an early prediction system. In this paper, the Air Quality Index (AQI) is predicted using Machine Learning Techniques.Various models including Linear regression, Decision tree, Random forest and XGBoost were implemented and compared. The dataset applied in this study contains common air pollutants such as PM2 5, PM10, CO, NO?, SO? and O? and weather conditions (temperature, humidity and wind speed). They clean and prepare the data properly before applying the models to enhance accuracy. Models performance evaluated using MAE, RMSE, and R² score. Upon testing different models, Random Forest and XGBoost perform better than others with XGBoost performing the best. The system established in this research could, Hydratil expect, be of assistance to warn air contamination early. Information resulting where the public and government can take action before it leads to big losses. In conclusion, this paper illustrates a practical application of Machine Learning for tackling environmental issues.

Introduction

Air pollution is a growing global concern driven by urbanization, industrial emissions, and increasing vehicle usage. Harmful pollutants such as PM2.5, PM10, NO?, SO?, CO, and O? significantly impact human health, leading to respiratory and cardiovascular diseases. Although air quality monitoring systems provide real-time data, they lack predictive capability, making it difficult to take preventive action before pollution peaks occur. To address this limitation, the study explores Machine Learning (ML) techniques for forecasting air quality and enabling early warning systems.

The literature review shows a shift from traditional statistical models like ARIMA to more advanced ML methods due to better performance on large datasets. Ensemble models such as Random Forest and XGBoost consistently outperform simpler models by improving accuracy and reducing errors. Deep learning methods like LSTM and CNN are also being explored for capturing temporal and spatial pollution patterns, though they require higher computational resources. Among all approaches, ensemble learning is considered the most reliable for AQI prediction.

The proposed system uses a layered architecture involving data collection from sensors and weather sources, preprocessing (cleaning and normalization), machine learning model application, and user-facing output through dashboards or alerts. Key features include pollutant levels and meteorological factors such as temperature, humidity, wind speed, and rainfall. Model performance is evaluated using MAE, RMSE, and R² score.

Experimental results show that XGBoost (93% accuracy) and Random Forest (90% accuracy) outperform Decision Tree, Linear Regression, and SVM. Feature analysis reveals that PM2.5 and PM10 are the most important predictors, with weather variables also contributing significantly. Overall, the study concludes that ensemble-based ML models provide highly accurate and reliable air quality predictions, making them suitable for real-time forecasting systems and early pollution warning applications.

Conclusion

This study has shown a good way to use Machine Learning to predict the Air Quality Index (AQI). Air pollution is a big problem that harms both people and the environment. We need systems that can predict pollution levels ahead of time. Traditional methods of monitoring are not very useful because they only give you information about the present and can\'t predict what will happen in the future.Various Machine Learning models, including Linear Regression, Decision Tree, Random Forest, and XGBoost, were utilised and evaluated to address this issue. The study\'s findings indicate that ensemble methods, particularly Random Forest and XGBoost, yield superior accuracy and enhanced performance compared to alternative models. XGBoost was found to be the best and most reliable way to predict air quality. The research demonstrates that using both pollution measurements and meteorological data improves their prediction accuracy. The results of the study depend on two essential processes which are data preprocessing and feature selection. The model will achieve better performance in the future through deep learning methods, which require larger datasets and real-time satellite data for precise wide-area predictions.

References

[1] Abuouelezz, W., Ali, N., Aung, Z., et al.. Exploring PM2.5 and PM10 ML forecasting models: a comparative study in the UAE. (Abuouelezz et al., 2025) [2] Ayus, I., Natarajan, N., & Gupta, D.. Comparison of machine learning and deep learning techniques for the prediction of air pollution: a case study from China. (Ayus et al., 2023) [3] Chadalavada, S., Faust, O., Salvi, M., et al.. Application of artificial intelligence in air pollution monitoring and forecasting: A systematic review. (Chadalavada et al., 2024) [4] Dalk?l?ç, E., & Dursun, ?.. Air Quality Prediction Using Programming Language in Konya. (Dalk?l?ç& Dursun, 2025 [5] Garbagna, L., Saheer, L. B., &Oghaz, M. M.. AI-driven approaches for air pollution modelling: A comprehensive systematic review. (Garbagna et al., 2025) [6] Library Source. Air Pollution Prediction System Using Machine Learning.

Copyright

Copyright © 2026 Rohit Kumar Singh SINGH, Utkarsh Verma, Yashi Singh, Mr. Pranav Rai. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.