多源异构数据驱动的后疫情时期旅游需求预测方法研究  

Multi-source Heterogeneous Data-driven Tourism Demand Forecasting Approach amid Post-COVID-19 Era

作　　者：武静 赵二龙 孙少龙 汪寿阳[2,3,4] WU Jing;ZHAO Erlong;SUN Shaolong;WANG Shouyang(School of Management,Xi’an Jiaotong University,Xi’an 710049,China;Academy of Mathematics and Systems Science,Chinese Academy of Sciences,Beijing 100190,China;Center for Forecasting Science,Chinese Academy of Sciences,Beijing 100190,China;School of Entrepreneurship and Management,ShanghaiTech University,Shanghai 201210,China)

机构地区：[1]西安交通大学管理学院,西安710049 [2]中国科学院数学与系统科学研究院,北京100190 [3]中国科学院预测科学研究中心,北京100190 [4]上海科技大学创业与管理学院,上海201210

出　　处：《计量经济学报》2023年第2期350-366,共17页China Journal of Econometrics

基　　金：国家重点研发计划项目(2022YFF0903000);国家自然科学基金(72101197,71988101);陕西省软科学项目(2023-CX-RKX-081)。

摘　　要：准确的旅游需求预测对旅游业高质量发展有着重要的作用,特别是中短期客流量预测对旅游目的地的旅游资源调配及应急管理至关重要,新冠疫情的冲击导致旅游需求出现结构性变化,将对旅游需求预测预警提出新的挑战.本研究工作主要聚焦后疫情时期多源异构数据融合是否能够提升旅游需求预测方法的性能,为了探究该问题,本研究首先从旅游者生成的在线评论、旅游者关注的网络搜索数据和旅游目的地实时在线新闻数据中提取众多影响旅游需求波动的变量,并采用主题模型、情感分析及特征工程方法对其进行处理;其次,根据变量时间频度的不同采用混频建模的方法对变量进行融合处理得到预测的多模态融合特征;最后,基于多源异构数据融合驱动SARIMA-MIDAS预测方法对旅游需求进行建模预测.该研究工作主要以后疫情时期中国香港游客量建模预测为研究对象,实证结果揭示出本研究提出的多源异构数据融合驱动SARIMA-MIDAS预测方法在后疫情期间能够取得最佳的预测表现,因此,该研究的结果可为旅游需求预测提供一种新的解决方案,为相关政府机构和从业者提供决策支持.Accurate tourism demand forecasting plays an important role in high quality development of tourism industry,especially short-and medium-term tourist arrivals forecasting is crucial for the tourism resource allocation and emergency management,the COVID-19 epidemic has led to structural changes in the tourism demand,posing new challenges to tourism demand forecasting and warning.This study focuses on whether fusing multi-source heterogeneous data can improve the performance of tourism demand forecasting approach amid post COVID-19 era,in order to explore this problem,firstly,a large number of variables affecting the fluctuation of tourism demand are extracted from online reviews generated by tourists,internet search data of tourists’attention and real-time online news data of tourist destinations,and topic model,sentiment analysis and feature engineering methods are adopted to deal with them in this study;secondly,mixed data sampling model with different data frequencies is used to fuse the variables to get multimodal fusion characteristics for forecasting;finally,fusing multi-source heterogeneous data-driven SARIMA-MIDAS approach is proposed for tourism demand forecasting.The main object of this study is tourist arrivals forecasting of Hong Kong amid post COVID-19 era.The empirical results reveal that fusing multi-source heterogeneous data-driven SARIMA-MIDAS approach can achieve the best forecasting performance amid post COVID-19 era.Therefore,the results of this study can provide a new solution for tourism demand forecasting and decision support for relevant government agencies and practitioners.

关 键 词：旅游需求预测 多源异构数据 混频数据融合 用户生成内容 情感分析 

分 类 号：F592.7[经济管理—旅游管理] F224[经济管理—产业经济]