基于AB@G集成模型的九寨沟景区游客量预测研究

Research on Prediction of Tourists' Quantity in Jiuzhai Valley Based on AB@G Integration Model

生态景区旅游业的发展与生态环境保护之间的矛盾已成为景区管理最为关注的焦点,而景区游客量的预测是解决该矛盾的首要任务。文章遵循集成思想,以季节性ARMA模型、神经网络模型及组合模型为基础,采用GMDH非线性叠加的集成方法,构建了适用于线性与非线性交错复杂特点数据的AB@G集成预测模型,并以九寨沟景区为研究对象进行实证分析,证明了该模型在预测游客量上是有效的。

Balancing the development of eco-tourism and environmental and ecological protection in scenic areas is an increasingly important issue for tourism management and the tourism sector in general. Tourism forecasting is a primary mechanism to help address this potential conflict. Research confirms tourism forecasting, which is a key tool of tourism management, can assist effective environmental protection. This is particularly useful when uncertainty over tourist numbers challenges the management of scenic areas contributing to environmental pollution. Accurate predictions assist scenic area managers in making informed planning decisions and efficiently allocating a variety of resources. This promotes the sustainable development of scenic areas while ensuring economic benefits. In the future, accurate tourism forecasting will become increasingly important for the effective management and the sustainable development of scenic areas. This study examines ways to improve the accuracy of tourism forecasting. This enhancement will provide better foundation information for management decisions and allow the tourism sector to develop in tandem with other relevant industries. Tourism forecasting using a simple single method or a combined linearsuperposition method yields poor results. The complex nature of changing tourist numbers means prediction models that are linear and nonlinear, and which are mutually integrated, are somewhat deficient. The most efficient way to improve forecasting accuracy is to introduce artificial intelligence （AI） techniques to an integrative method. This solves linear and nonlinear issues which are interlaced within tourism predictions. This paper proposes an integrative forecasting model which can be used to accurately produce forecasts of tourist numbers visiting Jiuzhai Valley. According to system integration thinking the AB@ G model provides the best accuracy of all models when used to predict tourism figures. Tourists＇ quantity alters and these shifts have complex linear and nonlinear characteristics. The AB@ G model is built based on the seasonal autoregressive moving average （ARMA） model, the neural network model and their combined model. These are integrated by the group method of data handling （GMDH） using the nonlinear superposition method. Therefore, the AB @ G model combines the linear characteristics of the autoregressive integrated moving average （ARIMA） model and the nonlinear characteristics of the Back Propagation （ BP ） model. Further, the AB @ G model also includes GMDH technology. It is an integrative method with nonlinear superposition overcoming defects of simple addition and improving prediction accuracy. An empirical analysis of tourism development and management in Jiuzhai Valley examines and clarifies the advantages of the AB@ G model. Results indicate the AB@ G model has a lower prediction error rate than other models. The average prediction error results of the AB@ G model were 3.21%. This compares with an average prediction error of 13.3% for the seasonal ARMA model, 11.8% for the neural network model and 10.8% for the combined ARMA and neural network model. Therefore, the AB @ G model is a valid tourism forecasting tool. Further, this paper draws the following conclusions： first, the performance of single models （linear and nonlinear） is not ideal; second, the combined ARIMA-BP model can improve forecasting performance, but this model is not optimal; third, the AI techniques, integrative method （GMDH） can effectively reduce prediction errors. Therefore, it can be used as an effective tool to forecast tourist numbers in the Jiuzhai Valley scenic area.