基于Elman神经网络的我国入境游客量动态预测

Prediction of Dynamic Elman Neural Network Based on Amount of Chinese Inbound Tourists

为了更好地拟合我国入境游客量系统动态和时变特性,对我国入境旅游游客量进行科学准确的预测,将Elman网络引入我国入境游客量预测,提出基于Elman神经网络的入境游客量预测模型。以国家统计局公布的最近30个月我国入境旅游游客量月度数据为研究对象,前27个月数据用于建立训练网络,后3个月数据用于检验模型,模型预测最大误差为2.928%,最小误差为1.492%,平均误差为2.19%,模型预测效果与实际非常接近。结果表明:模型能更好地反应我国入境游客量系统的非线性、动态性、时变性等特征,为我国入境旅游游客量预测提供了一种较可靠的途径。

In order to carry on Chinese inbound tourists quantity forecast, this paper proposed an ARIMA model and RBF neural network algo- rithm. In 2009 January to 2014 April, China inbound tourism tourist amount monthly data as the research object, firstly used the ARIMA prediction of China inbound tourists to compute the residual. Then using the RBF neural network forecasted the residual and the ARIMA prediction results of cor- rection. The methods were used the China inbound tourism tourist quantity prediction. The results showed that the ARIMA model was modified by using the RBF neural network, the linear fitting algorithm and nonlinear fitting algorithm were combined to predict China inbound tourists quantity was a more reliable algorithm.