基于分解集成的航空货运需求区间预测研究.

Air cargo is an important strategic resource of country and plays an indispensable role in domestic and international trade. Scientific forecasting of air cargo demand is an important basis for airlines to make infrastructure planning and overall investment decisions. Aiming at the uncertainty of air cargo volume data, this paper introduced Bootstrap method for uncertainty estimation and proposed an interval prediction method based on decomposition integration from the practical needs. Specifically, this paper decomposed the historical data by Seasonal and Trend Decomposition using Loess (STL) method firstly, then forecasted the trend and seasonal components by Support Vector Regression (SVR) and Seasonal Autoregressive Integrated Moving Average (SARIMA), respectively. Thirdly, this paper extracted and resampled the white noise component by Bootstrap method. Finally, the prediction results were integrated and reconstructed with the processed white noise to quantify uncertainty using quantile construction intervals. The experimental results of cargo data from two hub airports in China show that the constructed interval can effectively quantify the uncertainty in combination with the predicted results, which provides a novel research idea for probabilistic interval prediction. [ABSTRACT FROM AUTHOR]