Efficient goods inspection demand at ports: a comparative forecasting approach

• Author: María Cerbán‐Jiménez,María‐Jesús Jiménez‐Come,Jose‐Antonio Moscoso‐López,Juan‐Jesús Ruiz‐Aguilar,Ignacio Turias
• DOI: http://doi.org/10.1111/itor.12397
• Date: 01 September 2019
• Published date: 01 September 2019

Intl. Trans. in Op. Res. 26 (2019) 1906–1934

DOI: 10.1111/itor.12397

INTERNATIONAL

TRANSACTIONS

IN OPERATIONAL

RESEARCH

Efficient goods inspection demand at ports: a comparative

forecasting approach

Juan-Jes´

us Ruiz-Aguilara, Ignacio Turiasb, Jose-Antonio Moscoso-L´

opeza,

Mar´

ıa-Jes´

us Jim´

enez-Comeaand Mar´

ıa Cerb´

an-Jim´

enezc

aDepartment of Industrial Engineering and Civil Engineering, University of C´

adiz, Spain

bDepartment of Computer Engineering, University of C´

adiz, Spain

cDepartment of the General Economy, Universityof C ´

adiz, Spain

E-mail: juanjesus.ruiz@uca.es [Ruiz-Aguilar]; ignacio.turias@uca.es [Turias];

joseantonio.moscoso@uca.es [Moscoso-L´

opez]; mariajesus.come@uca.es [Jim´

enez-Come];

mariadelmar.cerban@uca.es [Cerb´

an-Jim´

enez]

Received 18 January 2016; received in revised form 30 October 2016; accepted 13 January 2017

Abstract

A high number of freight inspections carried out at Border Inspection Posts (BIPs) of ports could lead to

significant time delays and congestion problems within the port system, decreasing the efficiencyof the port.

Therefore,this work is focused on achieving the most accurate prediction of the dailynumber of goods subject

to inspection at BIPs. Five prediction methods were used for this aim: multiple linear regression, seasonal

autoregressive integrated movingaverage, generalized autoregressive conditional heteroskedasticity, artificial

neural networks, and support vector regression models. Several nonlinear tests were used to study the nature

of the time series and the best method was obtained by the comparison of the prediction results based on

performance indexes that provide the goodness-of-fit. The result of this study may become a supporting tool

for the prediction of the number of goods subject to inspection in BIPs of other international seaports or

airports.

Keywords: inspection forecasting; seasonal ARIMA; GARCH; artificial neural networks; support vector regression;

decision support systems

1. Introduction

A vast increase in global trade among Asia, America, and Europe (mainly maritime transport)

has given rise to a strong impact in supply chain operations. This has caused an increase in traffic

congestion and time delays in freight transport (Ke et al., 2012). Nowadays, congestion problems

are found in several important ports of the world, and some of their effects include rise in cost and

fall in competitiveness.One of the main consequences of congestion in these systems is the diversion

C

2017 The Authors.

International Transactionsin Operational Research C

2017 International Federation ofOperational Research Societies

Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA02148,

USA.

J.-J. Ruiz-Aguilar et al. / Intl. Trans.in Op. Res. 26 (2019) 1906–1934 1907

of trade to some other more competitive ports (Fan et al., 2012). Besides, this fact becomes one of

the stronger reasons for the application of traffic flow prediction procedures that anticipate possible

saturation situations.

In the multimodal/intermodal chains, seaports are the connection points between maritime and

land transportation systems.The goods transported in commercial vehicles or containers are subject

to a number of operations that may affect the efficiency and cost of transport. As ports constitute

the main continental gateways in general and for Europe and Spain in particular, custom, border,

and veterinary controls constitute an integral and critical part of the transfer process that goods

must go through. These activities may add significant complexity and delay to port operations,

especially in the actual context of high security.

Checks must be carried out at a Border Inspection Post (BIP). A good that enters the BIP

can suffer three types of inspections: documentary check, identity check, and/or physical check.

Therefore, the inspection causes a delay in the flow of goods throughthe supply chain and this is the

main goal of this research. The different situations that can be given at a BIP, where a fixed number

of inspectors is usually assigned in each shift, together with the increase of world merchandise

trade giving rise to the saturation of the system in certain periods, could create serious congestion

problems or bottlenecks in the supply chain and ports. This congestion not only greatly impedes

operations at the marine terminals, but also affects roadway networks around marine terminals.

An optimal and seamless service at the BIP is a key quality indicator to assess the level of port

service. A wrong operation of a BIP will result in the escape of companies to ports with a faster

performance of their inspection posts and a more reliable supply chain. In order to avoid this

situation, a short-term prediction of daily inspections is necessary to anticipate peaks and make the

system more efficient. Long-term predictions of port operations are usually carried out, but short-

term predictions have the same importance (Peng and Chu, 2009). BIPs are facilities with a fixed

number of inspectors previously scheduled, regardless of the amount of goods that pass through

them. This can produce a bad planning of the human and material resources. The consequences

represent days with an excess of inspectors, generating poor management of the available resources;

or days in which the number of inspectors is insufficient, thereby generating congestion and delays

in the supply chain. A short-term forecasting of the number of inspection atBIPs may be a powerful

solution. In this case, a long-term prediction might be misleading, due to the different seasonality of

the data not only weekly but also monthly. The use of short-term prediction models can add value

to the supply chain management. This kind of inspection volume forecasting tool can be used by

the different port organizations and administrations as support decision aid for planning facilities

and resources dedicated to the port inspection system.

Determining whether the prediction is short or long term is not a decision to be taken lightly.

On the one hand, it depends on the sector or the type of industry that is focused in the study. On

the other hand, the length of a forecast depends on how rapidly the sector under study changes

and how susceptible it is to these changes. However, the distinction between short- and long-term

forecasts is not always defined clearly. Typically, short-term forecasts cover the immediate futureup

to approximately one or two years. Consequently, they are primarily used for hourly, daily, weekly,

or monthly prediction tasks. Due to this, they mainly focus on determining the delivery, personnel

and work schedules, production, transportation, in order to establish inventory levels, planning

human and material resources, or detect congestion or delays in a certain flow. Meanwhile, long-

term forecasts are used usually fora period longer than two years. This kind of prediction is used for

C

2017 The Authors.

International Transactionsin Operational Research C

2017 International Federation of OperationalResearch Societies