Forecasting Research Of Tourism Demand Under Regular Epidemic Prevention And Control

Tourism plays a pivotal role in promoting the high-quality development of the society,but it is highly susceptible to external events.Since the outbreak and spread of the novel coronavirus,the global tourism industry has faced unprecedented challenges.In the context of normal prevention and control,we need to rationally think about the degree,scope and ways of the impact of force majeure risks such as the epidemic on the tourism industry.Accurate tourism demand prediction can not only help personnel engaged in the tourism industry optimize resource allocation,but also effectively and scientifically assess the impact of major public health events such as the novel coronavirus outbreak on tourism travel activities,which is of great significance for scientifically preventing the spread of the epidemic and improving the safety of tourism travel.At present,the existing research literature on tourism demand forecasting,when investigating the impact of emergencies such as COVID-19 on the tourism industry,has not yet deeply explored the impact of COVID-19 on the characteristics of tourism demand data in the era of COVID-19.The problem of how to effectively extract and quantify such data characteristics and incorporate them into the forecasting task needs to be solved.To sum up,based on different theoretical advantages such as big data technology and deep learning,and the internal connection between tourism demand data and epidemic impact characteristics,this paper innovatively proposes a tourism demand prediction model based on the internal and external mining of epidemic impact.The model framework mainly includes:1)Based on the data characteristics of tourism demand itself,the epidemic impact quantification method was proposed,the epidemic impact characteristics were mined internally,and the core information extraction method was proposed to reduce data redundancy.2)The two-stage model training method was proposed to effectively transfer useful knowledge from historical data before the epidemic and help the training and prediction of data models after the epidemic.3)Considering the multi-destination travel behaviors of tourists during travel,relevant data of surrounding areas are included into the prediction task,and a feature region selection method is proposed to reduce the data dimension.4)Considering the increasing attention of tourists to travel restrictions under the epidemic situation,Internet search index was proposed to be incorporated into the model as an exogenous variable reflecting tourists’ attention.5)The Internet search index coding imaging method was proposed to extract the potential impact of the epidemic in Internet data through image feature conversion,so as to facilitate the subsequent feature learning.Finally,the empirical study based on the tourist number data set of Haikou,Sanya,Jiuzhaigou and Siguniang Mountain scenic spots showed that the tourism demand prediction model proposed in this paper based on the internal and external mining of the epidemic impact achieved the optimal prediction performance,especially the epidemic quantification method proposed under the background of normal epidemic prevention and control,greatly improved the prediction accuracy.The proposed model has positive significance in the field of tourism demand prediction under the disturbance of public health events such as the novel coronavirus outbreak,and provides strong empirical research support for the scientific decision-making of related tourism departments and practitioners.