Air pollution prediction with machine learning: a case study of Indian cities

Area of Science:

• Environmental Science
• Data Science
• Public Health

Background:

• Air quality is crucial for human survival, but industrial, transport, and domestic activities release hazardous pollutants.
• Effective air quality monitoring and prediction are vital, particularly in developing nations like India.

Purpose of the Study:

• To analyze and predict air quality in 23 Indian cities using six years of data.
• To identify key pollutants affecting the Air Quality Index (AQI) and hidden patterns in pollution data.
• To evaluate the efficiency of machine learning models for air quality prediction.

Main Methods:

• Dataset preprocessing and feature selection using correlation analysis.
• Exploratory data analysis to uncover patterns and influential pollutants.
• Application of five machine learning models (Gaussian Naive Bayes, Support Vector Machine, XGBoost, etc.) after addressing data imbalance with resampling.

Main Results:

• Exploratory analysis identified key pollutants impacting AQI.
• A notable decrease in pollutants was observed in 2020, the pandemic year.
• XGBoost model showed the highest linearity between predicted and actual air quality data, outperforming other models.

Conclusions:

• Machine learning offers an efficient alternative to traditional methods for air quality prediction.
• The XGBoost model is highly effective for predicting air quality in Indian cities.
• Air quality analysis revealed significant trends, including improvements during the 2020 pandemic.