管理评论

管理评论 ›› 2025, Vol. 37 ›› Issue (2): 139-151.

• 电子商务与信息管理 • 上一篇    

突发公共卫生事件冲击下考虑多源异构大数据的旅游需求可解释预测研究

曾宇容1,2, 吴彬溶3, 王林4, 张金隆4   

  1. 1. 湖北经济学院信息工程学院, 武汉 430205;
    2. 湖北省互联网金融信息工程技术研究中心, 武汉 430205;
    3. 河海大学商学院, 南京 211100;
    4. 华中科技大学管理学院, 武汉 430074
  • 收稿日期:2022-04-25 发布日期:2025-03-06
  • 作者简介:曾宇容，湖北经济学院信息工程学院教授，博士；吴彬溶(通讯作者)，河海大学商学院讲师，博士；王林，华中科技大学管理学院教授，博士生导师，博士；张金隆，华中科技大学管理学院教授，博士。
  • 基金资助:
    教育部人文社会科学研究规划项目(22YJA630003)。

Interpretable Tourism Demand Forecasting Considering Multi-source Heterogeneous Big Data under the Impact of Public Health Emergencies

Zeng Yurong1,2, Wu Binrong3, Wang Lin4, Zhang Jinlong4   

  1. 1. School of Information Engineering, Hubei University of Economics, Wuhan 430205;
    2. Hubei Internet Finance Information Engineering Technology Research Center, Wuhan 430205;
    3. Business School, Hohai University, Nanjing 211100;
    4. School of Management, Huazhong University of Science and Technology, Wuhan 430074
  • Received:2022-04-25 Published:2025-03-06

PDF (PC)

17

摘要： 本研究利用历史旅游流量数据,新冠病毒感染确诊人数数据,旅游相关和疫情相关的百度指数,天气、节假日数据，设计了考虑突发公共卫生事件冲击下的自然景区每日旅游需求量预测框架。将与疫情相关的搜索引擎数据引入到旅游需求预测中，并提出了ADE-TFT可解释旅游需求预测新模型，其中自适应差分进化算法(adaptive differential evolution,ADE)用来智能高效地优化时域融合变换器(temporal fusion transformers,TFT)的超参数。TFT是一种基于注意力的深度学习模型，它将高性能预测与对时间动态的可解释分析相结合，在预测研究中呈现了优异的性能。TFT模型产生了可解释的旅游需求预测输出，包括不同输入变量的重要性排序以及不同时间步长的注意力分析。可解释实验结果表明，疫情相关搜索引擎数据能够充分反映出新冠疫情期间游客对疫情的担忧程度，研究结果为突发公共卫生事件冲击下的旅游需求高精度预测提供了理论支持。

关键词: 旅游需求预测, 可解释性预测, 复合指数, 深度学习, 突发公共卫生事件

Abstract: This study designs a framework for predicting daily tourism demand in natural scenic areas considering the impact of public health emergencies using historical tourism traffic data, the number of confirmed COVID-19 infections, tourism-related and epidemic-related Baidu indices and weather and holiday data. Epidemic-related search engine data are introduced into tourism demand forecasting, and a new ADE-TFT interpretable tourism demand forecasting model is proposed, in which the Adaptive Differential Evolution (ADE) algorithm is used to intelligently and efficiently optimize the time domain temporal fusion transformer (TFT). TFT is an attention-based deep learning model that combines high-performance forecasting with interpretable analysis of temporal dynamics, presenting excellent performance in forecasting studies. The TFT model produces interpretable tourism demand forecasting outputs, including importance ranking of different input variables and attention analysis at different time steps. The interpretable experimental results show that the epidemic-related search engine data can fully reflect the level of tourists’ concerns about the epidemic during the during COVID-19 outbreak, and the findings provide theoretical support for high-precision prediction of tourism demand under the shock of public health emergencies.

Key words: tourism demand forecasting, interpretable forecasting, composite search index, deep learning, public health emergencies