Scholarlink

<p>This study introduces a novel approach in analyzing the determinants of tourism expenditure in Italy, with a particular focus on the differences between domestic and incoming tourism. The analysis is based on microdata from two surveys: ISTAT’s “Travels and Holidays” and the Bank of Italy’s “International Tourism”. Expenditure forecasts for both categories were calculated using the eXtreme Gradient Boosting (XGBoost) machine learning algorithm, ensuring high predictive accuracy. To identify the main variables influencing tourism expenditure, SHapley Additive exPlanations (SHAP) values were calculated, providing a deeper interpretation of the contribution of each one of them. The results highlight both differences and common points in the determinants driving domestic and incoming tourism expenditure, underscoring the need for targeted policy interventions to address the distinct characteristics of these two segments. The methodological integration of machine learning techniques and SHAP values offers a tool for understanding and forecasting tourism expenditure trends in Italy, with relevant implications for policymakers and industry stakeholders.</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

Explaining Tourism Expenditure Patterns in Italy: A Comparative Study of Domestic and Incoming Tourists Using XGBoost and SHAP

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • Fabrizio Antolini,
  • Samuele Cesarini,
  • Ivan Terraglia

摘要

This study introduces a novel approach in analyzing the determinants of tourism expenditure in Italy, with a particular focus on the differences between domestic and incoming tourism. The analysis is based on microdata from two surveys: ISTAT’s “Travels and Holidays” and the Bank of Italy’s “International Tourism”. Expenditure forecasts for both categories were calculated using the eXtreme Gradient Boosting (XGBoost) machine learning algorithm, ensuring high predictive accuracy. To identify the main variables influencing tourism expenditure, SHapley Additive exPlanations (SHAP) values were calculated, providing a deeper interpretation of the contribution of each one of them. The results highlight both differences and common points in the determinants driving domestic and incoming tourism expenditure, underscoring the need for targeted policy interventions to address the distinct characteristics of these two segments. The methodological integration of machine learning techniques and SHAP values offers a tool for understanding and forecasting tourism expenditure trends in Italy, with relevant implications for policymakers and industry stakeholders.