Research On Tourist Forecasting Of Scenic Area Based On The Spatial-temporal Differentiation Characteristics Of Network Search

With the advance of national economic income level,the increasing improvement of transportation facilities and the reform of vacation system,tourism has become one of the basic lifestyle,for people have stronger willingness and ability to afford tourism consumption.However,the expansion of tourism demand and the uneven space distribution of tourism resources lead to lots of overloading situation in some scenic areas.Therefore,it is significant for scenic areas to forecast their tourist volume scientifically.Moreover,the Internet technology has deeply influence the social life,there exists a closely relationship between tourist activity and network search data.Based all of these,this thesis focuses on the tourist forecasting in the near distance tourist market,taking Zijin Mountain scenic area in Nanjing as example,choosing 40 cities in the Yangtze River Delta as the tourist market.The Peak Index,Coefficient of Variation,Geographic Concentration Index,Grainger causality test,Vector Auto-regression model,Impulse response function,Auto-Regressive and Moving Average model are used to study spatial and temporal characteristics of the daily tourist network search data and optimize the daily tourist forecasting model for the scenic area.The results enrich the research on spatial and temporal characteristics of the tourist network search data and tourist forecasting,also provide the theoretical guidance for the scenic areas'management,promoting the healthy development of the scenic areas.Attributed to the above,this thesis draws the following results:(1)There is a different search preference when tourist city search the tourism information.The key factors of tourist network search are the information of residence,travel and traffic.Farther cities'tourists tend to concentrate on the partial keywords when they search tourism information,while the close cities'tourists tend to search various keywords.Moreover,the network search of food keywords shows a trend of reducing from coastal to inland,and the network search of residence keywords shows that farther cities are more compared with the close cities,and the network search of travel keywords shows a nuclear distribution in space,where the capital cities or main source of tourist market has a higher search volume.(2)As for the temporal distribution of the whole network search data,both inside province cities and outside province cities show the "double peaks" model in a year and has a weekly cycle in a month,but the peak appears in different time.The outside province cities' peak shows a regular change around the holiday,and the search volume is more in Tuesday and Wednesday,less in weekends.The inside province cities' peak appears in June and July,and the search volume is more in Friday and Saturday,less in Sunday to Wednesday.As for the space distribution of the whole network search data,the geography has a weak limit to network search compared with the tourist flow,which shows a nuclear distribution in space.(3)There is a close relationship between daily network search data and tourist arrival.On the one hand,daily network search data is a precursor of tourist arrival.On the other hand,some cities' tourist arrival can also be a precursor of network search data,whose tourist arrival and network search data are both high.Furthermore,the period of network search data to tourist arrival will increase as the traffic time was delay,but the boundary effect has a little influence in close city whose period of network search data shows an annular distribution in space.(4)Combined the network search data in tourist forecasting model can improve the timeliness and accuracy of the prediction.The tourist forecasting model based on the spatial and temporal distribution of network search data can provide the scenic area for specific management and reference suggestion.