Machine Learning Based Predictive Modelling of Vehicle CO2 Emissions: An Extensive Examination of Automobile Features

Abstract

There is a growing interest in discovering novel and inventive methods to decrease the emission of carbon dioxide (CO2) from diverse sources, mostly due to concerns about safeguarding the environment and mitigating the adverse effects of greenhouse gas emissions, especially in the automotive sector. This study article explores the fascinating potential of predicting the quantity of CO2 generated by autos using machine learning (ML) based methodologies. In order to create a system that can predict carbon dioxide emissions depending on a number of factors, we are investigating machine learning. Significant details about the car, such as its construction, kind, number of cylindrical objects, horsepower, gearbox type, fuel type, and fuel consumption, are included in the dataset that forms the basis of our investigation. This project's main objective is to create accurate and dependable prediction models for calculating vehicle-generated CO2 emissions. Legislators that deal with the environment, corporations, and consumers could find great value in these models. We use a range of machine learning (ML) techniques, such as ensemble techniques, neural networks, decision trees, and regression algorithms, to accomplish this goal. This study assesses the prediction accuracy, adaptability to different situations, and data utilisation of several techniques. To determine which predictive models are most suited for predicting CO2 emissions within the limitations of the given information, we employ a comprehensive evaluation process. Additionally, we look at how different feature subsets affect prediction accuracy, providing an indication of the relative significance of different vehicle parameters in relation to CO2 emissions. Our results advance the field of predictive emissions models and provide insight on the primary drivers of vehicle-related greenhouse gas emissions.