The forecasting model´s establish and analyze of the demand of traveling between mini-three links

The mini-three links has become the most important transportation mode for the Taiwan´s businessmen of traveling between Taiwan and China. Therefore, to develop the mini-three links´ forecasting model can provide the traveling information of mini-three links to the air carriers, it would be helpful for the air carriers to devise operation plan. This study will combine ARIMA model and backpropagation neural network (BNN) of Artificial intelligence, which just developed recent years, to apply to forecasting mini-three links´ traveling requirement and establish suitable forecasting model. Through mean square error (MSE) and mean absolute percentage error (MAPE) measure forecasting achievement found out that the result of combine ARIMA and BNN mode is superior to individual forecasting mode of ARIMA and BNN. As a result, to combine ARIMA and BNN mode to set the forecasting of traveling between mini-three links is very accurate.