Data Mining Of China-ASEAN Visitor Volume Based On Multi-predictor Parallel Combination

With the rising trend of Chinese national entertainment consumption levels,the ASEAN region has attracted a large number of Chinese tourists thanks to its rich tourism resources and geographical advantages.Visitor volume is an important indicator of tourism,quantitative research and forecast of ASEAN tourists will not only help ASEAN government agencies to control the development of tourism in a macro way,but also to build a blueprint for China-ASEAN tourism cooperation in advance.The existing literature on the forecasting of tourism in the ASEAN region mainly remains qualitative analysis,and quantitative research methods focus on econometrics and time series models.In recent years,technologies such as artificial intelligence and data mining have become increasingly mature and have achieved remarkable research results in many fields.Therefore,the quantitative study of ASEAN tourists based on machine learning algorithm has a certain theoretical significance and a wide range of practical value.This paper selected three popular tourist destinations in the ASEAN region as the research object.First of all,it analyzes the background of ASEAN tourists' research and the limitations of domestic and international tourists' predictions.Based on the attributes of visitors and data selection principles,this article determines the per capita GNP of the source country(China),the population of the source country,the per capita GNP of the destination country(Singapore,Malaysia,Thailand),the SPI(customer country CPI/destination CPI),and exchange rate as the impact factor of tourists and annual Chinese tourists in three countries as forecast targets.Secondly,this paper proposes a tourist quantity forecasting model based on random forest,BP neural network and PSO-SVR.It uses the experimental data to fit the model and conducts the forecast of tourist quantity.After that,it compares and analyzes the fitting results and the forecasting accuracy of three models using R2,MAE,MSE and MAPE indicators.Which show that the fitting correlation index of three models is about 90%,and the accuracy of PSO-SVR prediction is relatively good.Because the single predictor has different forecasting focuses and weak forecasting accuracy,the paper proposed a multi-predictor parallel combination model based on weighted allocation to optimize the accuracy of the visitor quantity forecasting.By comparing the predicted values of visitors with real values,the results show that the parallel combination model can achieve better prediction accuracy.Finally,on the basis of quantitative analysis of the quantity of tourists from three countries,in order to uncover the situation of joint rises and falls among the three countries,a research on tourism integration of ASEAN countries based on Apriori was proposed.Experiments show that the growth of tourists between the three countries in three country affects each other and there is weak competition,which can provide data reference for ASEAN tourism integration.Finally,the research results of this paper are applied to the ASEAN Ocean Big Data Platform Coastal Tourism Module to diversify the display of tourism data,achieve the function of forecasting the amount of tourists,and provide users with high-quality decision support services.