Scholarlink

<p>Urban technological innovation (UTI) is crucial for sustainable development, highlighting the importance of understanding its relationships with carbon emission intensity (CEI). Existing research primarily addresses the linear associations of individual UTI variables with CEI, neglecting complex relationships between multiple variables. This study addresses these gaps by analyzing panel data from Chinese prefecture-level cities and developing an explainable artificial intelligence (XAI) model, which integrates a Bayesian-optimized machine learning (ML) model with SHAP (Shapley Additive Explanations) to quantitatively capture the non-linear relationships and interaction effects of multiple UTI variables on CEI. The findings demonstrate that (1) the Bayesian-optimized LGBM model outperforms other models, exhibiting high precision in capturing complex relationships between multiple UTI variables and CEI; (2) the non-linear contributions of UTI variables to CEI are quantified, identifying negative and positive threshold points; and (3) interactive effects among UTI variables are examined, with inverted U-shaped, inverted U-shaped, X-shaped, and fluctuating trends are identified. These findings provide valuable threshold references for policymakers to enhance urban CEI governance while offering a theoretical foundation for aligning technological innovation with low-carbon transition strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Towards sustainable urban development: Exploring the association between technological innovation and carbon emission intensity in China

  • 期刊论文

  • 发表日期：

    2025年8月31日

查看全文
  • Wentao Wang,
  • Dezhi Li,
  • Yang Wang,
  • Lingli Li,
  • Lugang Yu,
  • Shenghua Zhou,
  • Linfeng Zhang

摘要

Urban technological innovation (UTI) is crucial for sustainable development, highlighting the importance of understanding its relationships with carbon emission intensity (CEI). Existing research primarily addresses the linear associations of individual UTI variables with CEI, neglecting complex relationships between multiple variables. This study addresses these gaps by analyzing panel data from Chinese prefecture-level cities and developing an explainable artificial intelligence (XAI) model, which integrates a Bayesian-optimized machine learning (ML) model with SHAP (Shapley Additive Explanations) to quantitatively capture the non-linear relationships and interaction effects of multiple UTI variables on CEI. The findings demonstrate that (1) the Bayesian-optimized LGBM model outperforms other models, exhibiting high precision in capturing complex relationships between multiple UTI variables and CEI; (2) the non-linear contributions of UTI variables to CEI are quantified, identifying negative and positive threshold points; and (3) interactive effects among UTI variables are examined, with inverted U-shaped, inverted U-shaped, X-shaped, and fluctuating trends are identified. These findings provide valuable threshold references for policymakers to enhance urban CEI governance while offering a theoretical foundation for aligning technological innovation with low-carbon transition strategies.