Paper Key : IRJ************261
Author: Sangamesh M D
Date Published: 05 Jul 2024
Abstract
This project aims to predict CO2 emissions from vehicles using machine learning techniques applied to a dataset encompassing vehicle characteristics such as make, engine size, transmission type, and fuel consumption metrics. The primary objective is to develop an accurate regression model capable of forecasting CO2 emissions, crucial for understanding environmental impact and ensuring compliance with emissions regulations across diverse vehicle types. The workflow involves comprehensive data preprocessing to handle missing values and categorize variables, followed by exploratory data analysis to uncover correlations and insights. The chosen model, Linear Regression, is trained and assessed using metrics such as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). The model is serialized using pickle for seamless integration into a Flask-based web application, facilitating real-time predictions. By providing stakeholders with a tool to estimate and mitigate vehicle emissions effectively, this project contributes to sustainable automotive practices and regulatory transparency.