The Forecast Of The Daily Tourist Flow Volume In The Mountainous Scenic Spot Based On The Internet Searching Index

With the rapid development of economy,tourism has become one of the fastest growing industries.The rapid growth of tourism not only promotes the development of regional economy,but also brings a series of challenges such as management difficulties to the management of tourist attractions.However,due to its unique geographical features,the daily management of mountain scenic areas is more difficult.Therefore,in view of the effective scheduling of scenic resources,the accurate prediction of daily tourist flow in tomorrow and the day after tomorrow is of great significance to the management of mountain scenic areas,and can provide information support for the managers of mountain scenic areas in their daily management decisions.In this dissertation,Huangshan scenic spot is selected as a case study,and the prediction of tomorrow and the day after tomorrow tourist flow in Huangshan scenic spot is studied in combination with the demand of scenic spot management.And according to the different characteristics of tourist flow in different periods,the tourist flow can be divided into two types: ordinary day tourist flow and holiday tourist flow.Different prediction models are established to predict the two types of tourist flow.The main research work of this dissertation are as follows:(1)A method of DBN-PSOBP daily tourist flow prediction based on network search is proposed.Aiming at the complex and nonlinear characteristics of daily tourist flow,a correction model of tourist flow prediction error based on deep belief network and BP algorithm optimized by particle swarm optimization is established.Considering that there is a certain correlation between the search behavior of the relevant information of the scenic spot and the tourist flow of the scenic spot,the network search index is added into the selection of the input variables,the model first uses DBN to predict the daily tourist flow,and then uses PSO-BP model to correct the prediction error of DBN to get the final prediction result.The original model without network search index and the index model with network search index are established.Experiments show that the model has higher fitting degree and prediction accuracy for the trend of tourist flow change after adding network search index.(2)An optimized SVM holiday tourist flow prediction method based on network search is proposed.Aiming at the characteristics of strong nonlinearity and small sample size of holiday tourist flow,a prediction method of support vector machine(SVM)based on particle swarm optimization algorithm with position disturbance and adaptive inertia weight(PAPSO)is proposed.The PAPSO algorithm is used to optimize the parameters of SVM,and the method is compared with PSO-SVM,GPSO-SVM and EDPSO-SVM.The experimental results show that the prediction error rate of PAPSO-SVM is lower and the accuracy is higher.