Abstract:  Forecast of empty container throughput,especially of the recent monthlythroughput,is an important reference for dispatch and allocation of empty containers. An innovative statistical method is introduced in this paper.It takes the Hong Kong Port as an example, builds a semi-parametric linearregression model to forecast the recent monthly throughput of empty containersbased on some relative statistical data from 2001to2008. By comparing the forecast resultswith the actual data in 2009, the semi-parametriclinear regression isprovedto be highly accurate, which demonstratesthat the semi-parametric linear regression can be used to forecast the short-termmonthly throughput of empty containerswith appreciated high precision. This method can also be further applied inmanyother fields. 

Keywords: empty container, throughput, semi-parametriclinear regression, seasonal factor

1. Introduction 

Withthe increasing development of economic globalization, the foreign trade hasbeen keeping a high growing rate, which promotes the container transportationaround the world. Meanwhile, economydevelopment and foreign trade imbalance result in a largeflow of empty containers. According to thestatistics in 2007, about 50 million of empty containers were shipped betweenthe ports around the world, which is almost 20% of the total annual amount ofthe shipped containers. The matter of empty containers is more serious inChinese ports than in the other ports of the world since the quite imbalanceexport and import structure in China. For example, the throughput of emptycontainers in Shanghai Port was obviously rocketed up from 550 thousand TEU in1997 to 7,760 thousand TEU in 2007. It was increased by 12.95 times, muchgreater than 9.34 times, the increase rate of total container throughput inShanghai Port. Accordingly, thepercentage of empty containers was also increased from 21.7% in 1997 to 29.3%in 2007 as shown in Fig. 1.