Forecasting CO₂ emissions of fuel vehicles for an ecological world using ensemble learning, machine learning, and...

Area of Science:

• Environmental Science
• Computer Science
• Data Science

Background:

• Rising carbon dioxide (CO2) emissions from vehicles contribute to global warming and climate change.
• Accurate estimation and reduction of vehicle CO2 emissions are vital for environmental sustainability.
• Addressing greenhouse gas emissions is a critical global concern.

Purpose of the Study:

• To evaluate and predict CO2 emissions from fuel vehicles using various machine learning algorithms.
• To compare the performance of machine learning, ensemble learning, and deep learning paradigms for CO2 emission prediction.
• To identify the most effective algorithms for accurate and efficient CO2 emission forecasting.

Main Methods:

• A comparative regression analysis was conducted on 18 different regression algorithms.
• Algorithms from machine learning, ensemble learning, and deep learning were employed.
• Performance was assessed using R², Adjusted R², RMSE, and runtime metrics.

Main Results:

• Ensemble learning methods demonstrated superior prediction accuracy and lower error rates.
• Extreme Gradient Boosting (XGB), Random Forest, and Light Gradient-Boosting Machine (LGBM) showed high R² and low RMSE.
• Deep learning models (CNN, DNN, GRU) achieved high R² but required more training time and computational resources.

Conclusions:

• Ensemble learning algorithms are the most effective for predicting vehicle CO2 emissions.
• The study provides valuable insights for stakeholders aiming for environmental sustainability.
• Findings support the development of strategies to reduce greenhouse gas emissions from transportation.