XAI-driven assessment of urban circular carbon economy: Using China's pilot cities as a case study

Abstract

The confluence of circular economy and carbon emission mitigation strategies has catalyzed the emergence of a concept called the Circular Carbon Economy (CCE). This study introduces a novel CCE evaluation framework targeting urban areas and employs Explainable Artificial Intelligence (XAI) to bolster transparency and interpretability. Through the application of big data mining techniques, key indicators were extracted from literature and policy documents to construct a multi-dimensional CCE evaluation system. This system was integrated with the Explainable Boosting Machine (EBM) model-a key XAI technique-to resolve the precision-interpretability trade-off prevalent in traditional models. The EBM-based framework achieves black-box-level accuracy while providing transparent decision pathways through feature importance quantification. Empirical validation with China's Circular Economy pilot cities demonstrates its applicability, offering a transferable paradigm for global urban CCE assessments. Our findings show that industrial solid waste utilization and the promotion of green buildings are the most critical drivers for advancing CCE in urban areas. The city rankings generated by the EBM model bolster the scientific validity and reliability of our framework. Moreover, the study uncovers significant regional disparities, with cities located in eastern and central China generally exhibiting higher levels of CCE development compared to their western and northeastern counterparts. Based on these insights, we propose three key recommendations-targeted development strategies, differentiated evaluation frameworks, and balanced regional development-to lay the theoretical groundwork and offer practical guidance for urban-level CCE development.