A Study on Carbon Footprint Tracking Method Based on K-means++ Clustering in Regional Scenarios

This study presents a concise and interpretable carbon footprint tracking method tailored for low-carbon parks and campuses. By integrating K-means++ clustering, regression analysis, and Shapley value interpretation, the framework enables real-time scenario classification, emission anomaly detection, and source attribution. Multi-dimensional features—including energy system configurations, environmental conditions, and user behaviors—are used to cluster parks into typical emission scenarios, each with a defined regression score threshold. When a new scenario exceeds this threshold, Shapley analysis identifies the key contributing factors. A case study of a smart office park confirms the model's effectiveness in pinpointing carbon-intensive behaviors, such as reduced solar radiation and aggressive air-conditioning usage, offering practical guidance for targeted energy optimization.