Towards NEOM: Predicting Carbon Emission Using Machine Learning Approaches
Hossam Meshref, Ammar Nasser Alqarni, Ahmed Y Mobarki, Fahad Almalki, Hasan Alruqi
Abstract
This research explores the potential to reduce vehicle carbon emissions to support NEOM's goal as well as other KSA cities to reach zero emissions, focusing on the relationship between vehicles' specifications and emissions. By analyzing the “Fuel Consumption” dataset from Kaggle, this research offers insights for achieving environmental sustainability. It reviews both national and international literature that used different machine learning techniques to predict CO2 emissions. Data exploration results reveal significant correlations suggesting, for example, that vehicles with better fuel efficiency, higher mpg, tend to have lower emissions. The best prediction accuracy results, at 99.3%, was achieved by the LogitBoost Ensemble technique and interpreted by the Decision Tree technique, at 99.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> accuracy. As well, a few recommendations were suggested to policy makers'. We believe that the proposed findings underscore the designed models' effectiveness in guiding policy makers informed decisions toward NEOM's as well as other KSA cities sustainability ambitions.