Analysis and Prediction of Individual Carbon Footprints: An Integrated Machine Learning and Web Application Approach for Sustainable Behavioural Change

Climate change has accelerated due to unprecedented increases in greenhouse gas emissions, driven by both industrial activities and individual behaviour. In response, there is a growing need for accurate, personalized carbon footprint assessment tools that can empower users to adopt sustainable practices. In this work, we present an integrated framework combining robust data preprocessing, advanced ensemble machine learning techniques, and an interactive web application to predict individual carbon footprints. Our system utilizes a multi-dimensional dataset—including transportation, dietary choices, household energy consumption, and social activities—to train a series of predictive models such as Linear Regression, Decision Trees, Random Forests, Support Vector Regression, XGBoost, and CatBoost. The CatBoost Regressor achieved superior performance (MAE = 68.64, RMSE = 95.49, R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.991). Comprehensive feature analysis reveals that transportation habits and energy usage are the predominant drivers of emissions. Our approach is contextualized within a broad literature review that covers evaluations of carbon footprint calculators, ICT-enabled reduction strategies, advanced predictive and retrieval methods, global emission analyses, reviews on ICT and sustainability, usability studies, and further project and code insights. The developed system not only provides accurate predictions but also offers actionable sustainability recommendations. Future enhancements include incorporating regional emission factors and real-time data integration.