Explainable AI analysis for smog rating prediction.

Area of Science:

• Environmental Science
• Computer Science
• Data Science

Background:

• Smog significantly impacts human health and the environment.
• Vehicles are a major collective contributor to smog formation.
• Quantifying individual vehicle smog impact is challenging but crucial.

Purpose of the Study:

• To develop a machine learning model for predicting individual vehicle smog contributions.
• To classify vehicles based on their smog impact using a 1-8 rating scale.
• To leverage explainable AI for actionable insights into vehicle emissions.

Main Methods:

• Utilized a dataset including vehicle model, year, city fuel consumption, and fuel type.
• Employed Random Forest and Explainable Boosting Classifier models.
• Applied SMOTE (Synthetic Minority Oversampling Technique) for data balancing.

Main Results:

• Achieved 86% accuracy in predicting vehicle smog contribution.
• Reported Mean Squared Error of 0.2269 and R-squared of 0.9624.
• Incorporated explainable AI techniques for model interpretability.

Conclusions:

• The proposed machine learning approach effectively predicts vehicle smog impact.
• Results outperform previous studies, offering timely and relevant insights.
• This research is a significant step towards mitigating vehicle-related air pollution.