Total Domestic Tourism Consumption Forecast Based On ARIMA And ARIMAX Combined Models

This paper conducts a comprehensive and thorough analysis of the total domestic tourism consumption data from 2001 to 2019 to provide theoretical basis and technical support for the prediction of the trend of the tourism industryFirstly,76 quarterly data of total domestic tourism consumption and total domestic tourism population from 2001 to 2019 were selected respectively.The time series data treated by logarithm and first-order difference were found to be stable through graph test method and autocorrelation coefficient test method.Secondly,two single prediction models are selected for prediction.For the ARIMA model,the average absolute error is 6.7449%.However,its reference value is limited because it only contains its own historical information.In order to make the model more accurate,other auxiliary variables can be introduced appropriately.For the ARIMAX model,the average absolute error is 3.6872%.To make the model more accurate,a large amount of data is needed,but only 76 data can be collected and utilized at present.It can be seen that each single prediction method has its own shortcomings.Then,the sample information provided by two single prediction methods is considered synthetically.Based on the criterion of error square and minimum in combinatorial prediction,the combined model was constructed by inverse variance method,linear programming method and entropy weight method to solve the weight coefficient.The average absolute errors were 3.5807%,3.2825% and 4.2970% respectively.From the perspective of prediction error,time series combination models are obviously better than single models,because they integrated the dual characteristics of ARIMA model and ARIMAX model,which made up for the lack of auxiliary variables in ARIMA model and the lack of data in ARIMAX model.Among the three combinatorial models,the time series combinatorial model obtained by solving the weight coefficient with the linear programming method is better,and the prediction accuracy is improved by 0.2982% and 1.0145% respectively.Total domestic tourism consumption is expected to reach 640.034 billion yuan in 2020.It will continue to rise in the next few years,but the trend is weaker than that of 2015-2019.