近年来, 中国港口集装箱吞吐量增长迅速. 如何准确预测中国港口集装箱吞吐量是一个极其重要且具有挑战性的问题. 采用一种自适应方法来解决该问题, 即Yang (2004)提出的指数加权聚合预测(AFTER). 我们采用该方法将两种时间序列模型: SARIMA 和VAR 进行组合. 对中国七大港口的预测结果表明, AFTER方法比常用的简单平均预测具有优势, 它通常能以更高的频率自动设置更大的权重在更好的个体预测上.

Recently, Chinese ports container throughput increased rapidly. Accurate forecasting for container throughput of Chinese posts is an important and challengeable problem. We apply an adaptive approach --- Aggregated forecast through exponential reweighting (AFTER), introduced by Yang (2004) to this problem. Two classes of time series models (seasonal ARIMA model and vector autoregressive model) are combined. The results indicated that the AFTER approach has advantage over the common used simply average forecasting, and can automatically put a large weight on the better individual forecast with high frequency.