Research On Tourism Forecast Of Guangzhou Based On Search Indexes Of The Keywords

The tourism industry is a tertiary industry known as a smoke-free industry.Tourism is an information-intensive industry,its operations are deeply dependent on information.But for a long time,there have been problems such as asymmetry and diversification of tourism information.As a result,tourism resources have not been well developed and utilized.Nowadays the Internet has undergone rapid expansion.Different search engines have gathered a large amount of travel search information with higher timeliness.Combining the data on the Internet to predict the number of tourists is an improvement on traditional methods.Based on this,this research takes Guangzhou,a popular tourist destination city,as an example,introduces Baidu Index to reflect people's attention to Guangzhou tourism,and builds a combined model called AL-S-AFSA-LSSVR to precisely predict its monthly visitor numbers.Firstly,initial keywords are chosen by technical selection and other methods,then the lexicon is expanded to a collection of 387 keywords through a series of methods such as text mining.After screening out those not included in the Baidu Index and those with little date volume,a lexicon of 184 keywords is constructed,and their daily Baidu Index is extracted and converted to monthly indexes.Secondly,14 keywords are filtered through correlation analysis.These keywords have something in common: l(29)0 and 0.65 lr 3.Thirdly,this research uses Adaptive-Lasso method to further select 14 keywords,with the last 6variables chosen.Due to the strong seasonality of the tourist time series,2 seasonal related variables are introduced to the variable set together with the above-mentioned6 variables.Lastly,the AFSA-LSSVR model is constructed.117 sets of data are divided into two parts——one is the training sets with 97 sets,and the other is the test sets with 20 sets.Then the optimized and trained model is used to make predictions on the test sets.The final results shows that,compared with the real tourist numbers,the prediction has high accuracy.The MAE of the model is 12.64,MAPE is 2.70%and NSE is 0.9646.In order to evaluate the prediction accuracy of the model,this paper attempts to change the components of this combined model and make predictions on the same data set.Five indexes(RSR,MAPE,RMSPE,NSE and MAE)are introduced to evaluate them.Finally,this paper finds through the evaluation indexes,the AL-S-AFSA-LSSVR model has the highest prediction precision.On top of that,the DM test is carried out,and it proves that the AL-S-AFSA-LSSVR model is indeed better than most models at the significance level of 5%.Based on the analysis above,the model is expected to be used in practical work,and provide a reference for the planning,management,and scientific development of the tourism in cities.