Scholarlink

<p>The volatility of tourism demands under structural breaks pose significant challenges for destination management and policy planning. To address this critical issue, we developed a novel Hybrid Economic-Search Model (HESM) that synergistically integrates multidimensional data sources through an innovative two-stage forecasting architecture. Our framework uniquely combined composite diffusion index capturing macroeconomic regime shifts with a principal components-based Baidu Search Index reflecting real-time tourist intent. The model architecture employed a structured linear component to capture baseline relationships and interaction effects, followed by an XGBoost-based non-linear corrector to resolve residual patterns. Evaluated on the post-pandemic recovery period (2023–2024), HESM reduced the MAPE, MAE, and RMSE of the best-performing SARIMA benchmark by 27.07%, 26.62%, and 26.04%, respectively. Ablation studies further validated the individual and synergistic contributions of both economic and search data components. The model maintained robust performance across different market regimes, as confirmed through pre-pandemic validation. Our findings provide both methodological advancements and practical insights, demonstrating that hybrid frameworks leveraging multi-source data offer a paradigm shift in tourism forecasting accuracy under structural instability.</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

When simplicity fails: forecasting Mainland Chinese tourist arrivals in Macao during structural breaks with a hybrid economic-search model

  • 期刊论文

  • 发表日期：

    2026年2月23日

查看全文
  • Xiaoqing Luo

摘要

The volatility of tourism demands under structural breaks pose significant challenges for destination management and policy planning. To address this critical issue, we developed a novel Hybrid Economic-Search Model (HESM) that synergistically integrates multidimensional data sources through an innovative two-stage forecasting architecture. Our framework uniquely combined composite diffusion index capturing macroeconomic regime shifts with a principal components-based Baidu Search Index reflecting real-time tourist intent. The model architecture employed a structured linear component to capture baseline relationships and interaction effects, followed by an XGBoost-based non-linear corrector to resolve residual patterns. Evaluated on the post-pandemic recovery period (2023–2024), HESM reduced the MAPE, MAE, and RMSE of the best-performing SARIMA benchmark by 27.07%, 26.62%, and 26.04%, respectively. Ablation studies further validated the individual and synergistic contributions of both economic and search data components. The model maintained robust performance across different market regimes, as confirmed through pre-pandemic validation. Our findings provide both methodological advancements and practical insights, demonstrating that hybrid frameworks leveraging multi-source data offer a paradigm shift in tourism forecasting accuracy under structural instability.