Vehicle Emissions Classification using Feature Selection and Machine Learning for Accurate CO₂ Rating Predictions

Vehicle emissions are a major contributor to environmental pollution and climate change, posing significant risks to public health and global sustainability. Accurate prediction of CO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> ratings is essential for effective regulation and mitigation strategies. By integrating sophisticated feature selection techniques and classification algorithms, the research addresses critical challenges such as feature optimization and class imbalance. In this study a comprehensive suite of statistical and machine learning methods is employed to extract significant patterns from the dataset, resulting in improved predictive accuracy and efficiency by employing techniques such as SMOTE for data balancing and feature selection methods like PCA, Fisher Score, and Information Gain. Among the models evaluated, the Decision Tree classifier combined with Fisher Score achieved the highest accuracy of 93% and macro avg. f1-score of 89%, outperforming other models. The findings of this research contribute to the development of more robust and practical emission prediction models, offering valuable insights to mitigate environmental pollution and inform policy decisions.