Tourist Volume Forecast Of Chongqing Based On Baidu Search Index

When the network develops rapidly,consumers rely more on the Internet for their activities and decisions.The traditional offline consultation and other traditional channels have been replaced by search engines containing a large amount of important information.Therefore,to some extent,search keywords can represent the focus of consumers' attention,while search volume can represent the degree of attention.Among them,Baidu search engine has been occupying the largest market share in China,and the personal preferences,purchase needs,hot spots and other information of Internet users are stored in Baidu search.On the other hand,with the gradual improvement of people's living standards and the substantial increase of disposable income,people's demand for tourism is growing,which makes it the best choice for more people in their spare time.However,the traditional tourism decision-making mode can not meet the current situation of the surge of tourists.It has become a trend for consumers to collect tourism information through the search engine platform.Therefore,this paper takes Chongqing as the example,through the extraction and application of Baidu search data,analyzes the relationship between the data of Internet search keywords of consumer and the number of tourists,and establishes the corresponding statistical model to predict the number of tourists.The traditional research methods of tourism volume prediction have obvious lag and subjectivity.The network search data will be updated every day,and the daily value obtained has strong timeliness,which just avoids the lag of traditional methods.This paper studies through five steps:First,28 initial keywords are selected based on seven factors related to tourism by using the methods of literature and subjective word selection,and then the demand spectrum of Baidu Index and the home of stationmaster are used to expand the initial keywords to 105.Second,the monthly tourism volume of Chongqing and Baidu search index of each keyword are extracted.Through preprocessing,time difference correlation analysis and random forest importance ranking,23 final keywords were selected.Thirdly,Through PCA,23 keywords data were dimensionally reduced,and three prediction models of random forest,support vector machine and BP neural network are constructed respectively with the extracted five principal components and Chongqing tourist volume data.Fourth,contrsating the prediction effect of these models,the multiple regression comprehensive prediction model is established on the basis of the three single prediction models.Fifthly,the four models are evaluated according to the model comparison standard,and the comparison chart between the prediction data and the actual data is given,the fitting effect of the comprehensive prediction model is the best.There are main conclusions of this study:(1)The key words of scenic spots,such as Chongqing Yangtze River and Chongqing Chaotianmen cableway,are the most closely related to Chongqing's tourist volume,which is a major factor in the change of Chongqing's tourist volume and a huge driving force for the growth of tourist volume.(2)Internet search data can really reflect consumer decision-making,and it is feasible to use Internet search data to solve practical problems.It can effectively predict the number of tourists.(3)The average absolute error BP neural network prediction is better than the random forest prediction,but the practicability of the random forest is better than the BP neural network prediction.(4)Comparing the single models and multiple regression model,it is found that the multiple regression comprehensive prediction model is the best in both intuition and practicability,and the prediction effect has improved significantly on the basis of single models.