The Asian Journal of Shipping and Logistics 38 (2022) 222–233

Contents lists available at ScienceDirect

The Asian Journal of Shipping and Logistics

journal homepage: www.elsevier.com/locate/ajsl

Original article

The central tendency of the seaport-fulcrum supply chain risk in

Indonesia using a rough set ]]
]]]]]]
]]

⁎
Muhammad Reza Do. Bagus a,b, , Shinya Hanaoka a
a
Department of Transdisciplinary Science and Engineering, Tokyo Institute of Technology, Japan.
b
Department of Information System, College of Informatics and Computer Management Tidore Mandiri (STMIK Tidore Mandiri), Tidore, Indonesia.

a r t i cl e i nfo a bstr ac t

Article history: Seaports are crucial in modern global supply chain networks and systems, exerting a significant impact on
Received 17 January 2022 the ecological area's economy. Supply chain entities play a substantial integration role, and potential threats
Received in revised form 2 June 2022 at seaports harm chain continuity. This study proposes a rough set-based genetic algorithm, to investigate
Accepted 31 August 2022
the central tendency in seaport risk implied by supply chain threats through a questionnaire evaluation. We
also employ the risk score to observe the level of clarity in terms of risk probability, showing that the lower
Keywords:
the score an attribute obtains, the more likely it is that seaport risk implies supply chain disruption. We
Seaport-fulcrum supply chain risk
Risk assessment deploy 24 risk attributes, which threaten the proposed ten-dimensional factors, based on their risk scores.
Rough set theory The results show that the lack of storage risk planning, low punctuality of delivery goods, shortage of port
Genetic algorithm capacity, congestion in waterways, and the lack of distribution risk planning, are the "best five" of the
Central tendency analysis seaport-fulcrum supply chain risk, in the context of Indonesian seaport firms. These identified risk attri-
butes not only assist seaport managers to identify potential risk-associated deficiencies of supply chain
disruption but also enhance their ability to determine resilience to manage supply chain problems.
© 2022 The Authors. Production and hosting by Elsevier B.V. on behalf of The Korean Association of
Shipping and Logistics, Inc. This is an open access article under the CC BY-NC-ND license (http://creative-
commons.org/licenses/by-nc-nd/4.0/).

Introduction assessment needs to involve various elements related to the mar-

The exchanges and relationships between a seaport and its per-
ipherals generate globalization and supply chain continuity.
Seaports’ role in the global production and distribution system ac-
counts for more than 80% of international trade (Hall, 2007). As an
intersection between the worldwide mobility chain of goods and
people, ports have become critical to effectively and efficiently
evaluate as well as manage seaport-fulcrum supply chain risks,
protect the people and the environment, and maintain quality and
performance. Disruptive events at the seaport spread to various
seaport-fulcrum supply chain stakeholders and dimensions. The
identification of seaport risk involves understanding the supply
chain threats due to disruptive events. Correctly identifying those
risks contributes to the logistics industry by increasing seaport re-
silience and ensuring business sustainability. Therefore, the risk
⁎
Correspondence to: 2-12-1-I4-12, O-okayama, Meguro-ku, Tokyo, 152-8550, Japan
E-mail addresses: do.m.aa@m.titech.ac.jp,
dobagusmuhammadreza@gmail.com (M.R. Do. Bagus),
hanaoka@ide.titech.ac.jp (S. Hanaoka).
itime industry, such as ships, facilities, and other objects.
Enhancing the awareness of seaport risks and concerns helps
achieve port resilience. Although some studies (Dewi & Purnamasari,
2021; Esteban et al., 2020; Jiang, Li, & Shen, 2018; Loh, Zhou, Thai,
Wong, & Yuen, 2017; Morris, 2020; Weng, Liao, Wu, & Yang, 2020)
investigated such phenomena, research on the dynamics between
large numbers of seaport-fulcrum supply chain disruption factors
and the correlated risks is scant. Therefore, this study aims to pro-
vide an approach for identifying the central tendency in supply chain
risks and elaborating the impact of seaport risk on the threats to
supply chain continuity. To this end, it investigates the current
practices of seaport operations, where disruption management by
seaport firms significantly affects the seaport-fulcrum supply chain.
We select Indonesia as a case study. The data from Kementerian
Perhubungan R.I. (2020) shows at least 9755 cases of disruption
management. Typical causes of disruption are "disobedience" in
terms of operational rules, administrative regulations, and ministry
decrees; weakness in the control systems, such as accounting and
financial control; and policy. Both directly and indirectly, these fac-
tors relate to the export and import trade, as well as supply chain
continuity and accidents with victims (either infrastructure or
people). These phenomena reduce the seaport risk predictability.

https://doi.org/10.1016/j.ajsl.2022.08.003
2092-5212/© 2022 The Authors. Production and hosting by Elsevier B.V. on behalf of The Korean Association of Shipping and Logistics, Inc. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
M.R. Do. Bagus and S. Hanaoka
This study defines risk as the probability of a risk event in seaport
operation, multiplied by the impact of that risk (Aqlan & Lam, 2015).
Threats are situations that may trigger a hazardous source, generate
disruptive events, and raise the risk probability in the supply chain
(Singh, 2017). Thus, we define the seaport-fulcrum supply chain risk
as the potential effect of seaport risk on supply chain disruption.
We utilize a rough set-based genetic algorithm to reduce un-
certainty and deal with many risk factors. Rough set provides valu-
able tools for understanding data, and quantifying and handling
uncertainty, knowledge discovery, and vagueness in risk data. Wu,
Yue, Li, and Adjei (2004) stated the useful applications of rough set
theory as: pattern recognition and information processing, business
and finance, industry and environmental engineering, and intelligent
control systems. Moreover, the combination of rough set with other
methodology was also used in medical research (Sudha, 2017),
which integrated a rough set-based genetic algorithm with a neural
network to diagnose disease from clinical data sets. We use heuristic
information, such as the genetic algorithm, to determine the central
tendency of the risk factors and obtain a relatively minimal reduc-
tion among the seaport risk factors. Thus, the proposed rough set-
based genetic algorithm handles the complexity between the sea-
port risk factors and its supply chain entities.
Literature review
Many studies investigate the role of seaports in supply chains.
Some studies have proposed a comprehensive investigation of de-
velopments in seaport functions, based on three generations. For
example, Bichou and Gray (2005) show that seaports once provided
the convenience of cargo storage, but eventually, services such as
cargo distribution, packing, and processing have become crucial
seaport functions for conventional and bulk cargo in second-gen-
eration seaports. Loh and Thai (2015) examine the relationship be-
tween seaports and supply chains entering third-generation
seaports. They address the emerging concepts of cooperation and
information sharing—with seaports acting as facilitators—and tech-
nology playing a pivotal role in information exchanges between
different stakeholders. The three seaport generations have faced
changes in seaport ownership, port development strategy, and the
scope of port activities.
Seaports are crucial systems for coastal cities and global supply
chains. They are an attractive part of a city and facilitate the import,
export, and transportation of goods. Mokhtari, Ren, Roberts, and
Wang (2012) claim that these infrastructure systems may affect a
country's cost structure, industry competitiveness, and living stan-
dards. Seaports are equally crucial for trade networks, where
channel control and ownership may be determined or traded. In the
Indonesian context, an imbalance of cargo distribution, such as the
availability of infrastructure, shipping patterns, and supply and de-
mand of maritime transport including port connectivity, between
the western (developed economic region) and eastern areas (de-
veloping economic region) create a challenge in the seaport-fulcrum
supply chain risk disruption (Amin, Mulyati, Anggraini, and
Kusumastanto (2021). Rumaji and Adiliya (2019) addressed the issue
of high logistics costs and price disparity between both regions.
Some of the developed region, however, is still struggling with dwell
time, inefficient maritime security inspections at sea leading to
lower ship productivity, claims and contract cancellations, and
higher costs of voyages (Dewi & Purnamasari, 2021; Komarudin,
Reza, Moeis, & Rahmawan, 2017; Zaman, Vanany, &
Awaluddin, 2015).
Overall, seaports play a crucial role in different commercial ac-
tivities; however, they are highly complex systems, requiring a high
number of employees required, due owing to the widespread use of
personnel and technological equipment. For instance, port strikes in
Australia had left supply chain entities unable to fill orders, resulting
223
The Asian Journal of Shipping and Logistics 38 (2022) 222–233
in severe breach of contractual obligations (Loh et al., 2017). Cata-
strophic events due to natural disruption and pandemics may affect
all ports. The level and duration of the impacts depend on effective
protection measures (Notteboom, Pallis, & Rodrigue, 2021; Vous-
doukas et al., 2017). Additionally, seaports are particularly exposed
to extreme threats that may cause various risks, such as operational,
environmental or natural, security, technical, and organizational
risks (John et al., 2014).
In the supply chain risk assessment, scholars use the Analytical
Hierarchy Process (AHP) (Radivojević & Gajović, 2014), the fuzzy
comprehensive evaluation method (Aqlan & Lam, 2015), and various
other mainstream risk assessment methods. Some adopt relatively
novel evaluation methods. For example, Bogataj and Bogataj (2007)
use frequency and net present value analysis methods to evaluate
cost-associated risk factors. Klibi and Martel (2012) assess supply
chain network risk using Monte Carlo simulation. They indicate that
the combination of scenario analysis and Monte Carlo simulation
may be able to better evaluate the risk fluctuations associated with
supply chain network demand.
The above literature review indicates that seaports are increas-
ingly integrated into supply chain continuity, and thus, a disruptive
event originating from ports can harm these entities. However, many
supply chain risk identification and assessment models emphasize
the manufacturing sector and its enterprise without considering the
seaports’ significance in global supply chains. Further exploration
seems to be required to elaborate on the features of supply chain
threats, in particular, the presence of multiple seaport risks. These
deficiencies that directly affect seaport functionality and further
disrupt supply chain continuity are shown in Table 1. Additionally,
the hierarchical structure of the risk management model not only
includes supply chain operations risk focusing on the manufacturing
process but also covers the external risk. From a port enterprise’s
perspective, the operations and development of the seaport-fulcrum
supply chain are closely related to the import and export trade.
Hence, seaports face both external environmental and operational
risks. The proposed risk model is used to investigate the association
between the seaport risk and threats of supply chain disruption by
expert evaluation.
Materials and methods
Survey data
We employed an online questionnaire survey through face-to-
face interviews using a stratified random sampling technique with
Slovin’s formula. The period of data collection was January to August
2021. This study classifies the seaport-fulcrum supply chain stake-
holder into three main objects that significantly impact supply chain
issues: seaport-managers (10%), seaport-operators (40%), seaport-
users (50%). As a result, cargo owners, freight forwarders, ship
owners, and ship management companies are among the seaport-
users targeted.
A total of 153 data units were collected, in accordance with the
planned collection target of 150 data units. The demographics of
these respondents are presented in Table 2. Furthermore, the po-
tential of supply chain disruption is examined as a threat dimension
and to generate conditional risk factors evaluated by a five-level
ordinal scale, which indicates the risk implication in the supply
chain continuity. Next, a decisional factor is developed to assess this
term and determine the impact of seaport disruption on seaport-
fulcrum supply chain risk. Both risk factors are evaluated by the
seaport-fulcrum supply chain stakeholders. It adapts a rough set
model to find the central dependency in a large number of attributes
through a questionnaire survey. The proposed algorithm generates a
reduction attribute set that helps obtain a core attribute set. The set
of core attributes is crucial for understanding the center of the
M.R. Do. Bagus and S. Hanaoka The Asian Journal of Shipping and Logistics 38 (2022) 222–233

Table 1
The seaport-fulcrum supply chain risk events.

No. References Seaport Risk Events Threats to Supply Chain Disruption

1. Loh and Thai (2015);Bichou and Stakeholder coordination • Ownership profile and leasing structures of terminal operators experienced
Gray (2005) conflicts of interest, hindered the decision-making processes, and disintegrated
the supply chains.
2. Amin et al. (2021); Rumaji and Cargo throughput imbalance • The lower demand for containers in the developing regions of Indonesia leads
Adiliya (2019) to increased sea transportation costs, reduces port performance, and costs
maritime logistics.
• Lowers port performance including infrastructure availability, reduces the level
of competitiveness, and makes visible the bullwhip effect vulnerabilities in the
supply chain.
3. Dewi and Purnamasari Maritime security inspection • Causes operational delays, lowers ship productivity, increase voyage costs, and
(2021);Komarudin et al. (2017) increases exposure to liabilities under contracts of carriage.
• Too many agencies involved in maritime security with no clear division of
responsibilities leads to corruption.
4. Loh et al. (2017); Port strikes • Inability to fill orders, breach of contractual obligations, adverse effects in
manufacturing, retail, and food industries.
• Delayed, duplicated or lost shipments of supplies, delayed shipments to
customers, inventory build-up.
5. Notteboom et al. (2021); Vousdoukas COVID-19′s effect and natural • The lockdown affected most of the workforce and curtailed the industrial base
et al. (2017) disaster on the seaport operation between mid-January and early March 2020.
• A demand shock with backpropagation along supply chain.
• Last-mile vulnerabilities in distribution became visible because of the lower
availability of the workforce (e.g., absenteeism in trucking from major seaport).

Table 2
Demographic information of respondents.

No. Demographic factor Percentages

1. Gender Male 70%

Female 30%
2. Object of research Seaport-manager 11%
Seaport-operator 43%
Seaport-user 47%
3. Work duration Below 5 years 4%
Between 5 and 10 years 39%
Over 10 years 57%
4. Educational background Diploma 10%
Bachelor 61%
Master’s 27%
Doctoral 2%

seaport-focal supply chain risk tendency. Fig. 1 illustrates the clas-

sification process workflow.
The questionnaire presented the scale definitions before the
questions. Following the definition, the list of seaport risk and di-
mensional threat factors were provided in a table, with distinct
columns for respondents to insert their rating. The evaluations were
used as inputs for the proposed model.

Selection of dimensional threats and risk attributes

We mainly source the identification of dimensional threats and

risk attributes from accident reports in the literature. The 61 risk
attributes result from an extensive examination of the literature
(Section 2). We identify several indexes for capturing different per-
spectives of domain experts. They are categorized into dimensional
threat groups with the top event and risk attributes as conditional
factors, as shown in Table 3. We design this classification of threats
to reflect the responsiveness to the threat, in terms of the seaport Fig. 1. The classification process of attribute reduction.

supervision level, and define the scope of actions that may be an-
ticipated.
Basis of rough set
The rough set theory, initially introduced by Pawlak in 1985 (Wu
et al., 2004), is a mathematical approach for understanding and
manipulating imperfect knowledge. It helps deal with many risk
factors in risk assessment without losing classification capabilities
and attribute reduction. We use a genetic algorithm based on the
positive regions of a rough set to identify the core attribute and find
a relatively minimal reduction with reduced computational time.
A new data table is identified by the quadruple (4-tuples) S = {U,
A, Va, f}, where U is a finite set of objects (universe), A = C ∪ D; {a1, a2,
…, an} is a finite set of attributes (seaport risk), Va is the value set of
attribute a, where V = 1, 2,…, 5 indicates the highest to the lowest
evaluation, V = ∪a ∈ A Va, and f: U × A → V is a total function, such that f
(x,a) ∈ Va for each a ∈ A, and x ∈ U is called the information function.
A string vector describes each object x of U. Thus, the description of x

224
M.R. Do. Bagus and S. Hanaoka The Asian Journal of Shipping and Logistics 38 (2022) 222–233

Table 3
The seaport-fulcrum supply chain risk.

Dimensional threats Index Ai Conditional risk attributes Index aij References

Planning Process threats A1 Lack of seaport-enterprise strategic risk a11 Jiang et al. (2018);Loh et al. (2017).
Lack of berth risk planning a12
Lack of supply chain strategic risk planning a13
Lack of ship risk planning a14
Lack of handling process risk planning a15
Lack of storage risk planning a16
Lack of transfer risk planning a17
Lack of distribution risk planning a18
Deficiency of berth allocation risk planning a19
Infrastructure threats A2 Port equipment breakdown a21 Loh et al. (2017).
Inadequate port cargo handling equipment a22
Occupational accidents a23
Power outages a24
Breakdown of vessel traffic management system a25
Breakdown of port information system a26
Collisions in the waterway a27
Seaport Service Process threats A3 Congestion in the waterway a31 Jiang et al. (2018);Loh et al. (2017).
Congestion within terminals a32
Congestion at hinterland transfer a33
Less services calling at port a34
Less ship visits a35
Less load factors in captive cargo a36
Shortage of facilities or equipment a37
Shortage of port capacity a38
Shortage of IT and advanced technology a39
Distribution Process threats A4 Less timeliness of port departure and entry a41 Jiang et al. (2018);John et al. (2014);Loh
Low punctuality of delivery goods a42 et al. (2017).
Less timeliness of port customs clearance a43
Bad defect condition of goods a44
Low deviation time a45
Low efficiency of navigational services a46
Long time in feeder link a47
Less quality of logistics company a48
Relationship Process threats A5 Lack of member coordination a51 Jiang et al. (2018).
Member exit mechanism a52
Port labor strikes a53
Less motivation of member interest distribution a54
mechanism
Member information asymmetry a55
Nuclear-enterprise financial A6 Low revenue a61 Jiang et al. (2018).
threats High debt a62
Low-efficiency operation a63
Low growth development a64
Less cash flow a65
Less growth of domestic and international a66
macroeconomic operation
Monetary threats A7 Less efficient deviation cost a71 Kavirathna, Kawasaki, Hanaoka, and
Less efficient port cost a72 Matsuda (2018).
Less efficient cost in feeder link a73
Location threats A8 Short sailing time to the other hub ports a81 Kavirathna et al. (2018).
Less accessibility of hub port a82
Long connectivity of feeder markets a83
Security threats A9 International trade-war a91 Jiang et al. (2018);Loh et al. (2017).
War or terrorist attacks a92
Stowaway a93
Smuggling a94
Trafficking a95
Exchange rate a96
Environmental threats A10 Earthquake frequency a101 Jiang et al. (2018);Loh et al. (2017).
Pandemics/epidemics occurrence a102
Typhoon frequency a103
Increasing sea-level in the seaport a104
Increasing sedimentary level in the seaport a105
Decisional Factors D Implication of seaport risk to the potential threats of supply chain continuity.

is expressed in terms of the evaluation of the attributes from A. It
represents the available information about x, as:
Ades (x) = {f (x , a1), f (x, a2), …, f (x , an)}
An indiscernibility relation in U is every non-empty subset of
attributes B, denoted by IB, defined as follows:
IB = {f (x1, a) = f (x2, a), a B}
If (x1, x2) ∈ IB, objects x1 and x2 are B-indiscernible. The indis-
cernibility relation thus defined, is an equivalence relation (reflexive,
symmetric, and transitive). This relation also serves as a bridge for
(1) the proposed model, reducing the geometric increase of possible
profiles and determining the appropriate data presentation, con-
nection, and function form.
Let S be a data table, and X be a non-empty subset of U, and ∅ ‡ B
(2) ⊆ X. The set X may be characterized by two ordinary sets, the B-lower

225
M.R. Do. Bagus and S. Hanaoka
approximation of X (denoted by B (X )) and B-upper approximation of X
¯
(indicated by B¯ (X )) in S. These sets may be defined as follows:
B (X ) = {x U : IB (x) X}
¯ and design of the fitness function are considered first. The former
B¯ (X ) = {x U : IB (x) X }
The rough approximations obey the following basic laws: the
inclusion property: B̄ (X) ⊆ X ⊆ B̄ (X); the complementarity property:
B̄ (X) = U - B̄ (U – X). If an attribute from the subset B ⊆ A preserves the
indiscernibility relation RA, the attributes A–B are dispensable. D
totally depends on C only if I(C) ⊆ I(D), meaning that the partition by
C is finer than that generated by D. The concept of dependency
discussed above corresponds to that considered in relation to con-
ditional seaport risk datasets, where k = γ(C, D) is mathematically
defined as:
|POSC (D)|
(C , D) =
| U|
where
POSC (D) = C (X )
* |U|
X U / I (D)
POSC(D) is a positive region of the partition U/D with respect to C,
namely, the set of all elements of U that may be uniquely classified
into blocks of the partition U/D with respect to C.
Similarly, as the effect of the rough approximation law, the
conditional attribute set is C = {a1, a2, …, an} ∀ B ⊂ C, and the de-
pendency of the decision attribute on B is defined as:
|POSB (D)|
(B , D) =
| U|
To maintain the convergence speed and achieve the global op-
timum, while preserving the knowledge in the dataset, we propose a
preliminary step to determine the pre-classification model as a
significance attribute, where the conditional attribute subset is IB ⊂ C
∀ ai ∈ IB, and the significance of IB is defined as follows:
|POSC (D)| |POS(B ai) (D)|
Sai (B) =
|U|
The more B is reduced, the larger the average of the significance
attribute. This principle is reflected in the weight indicator in Eq. (9).
The theorem and proof from Eqs. (8) and (9) are shown in Appen-
dixes A and B, respectively.
|B|
i = 1 Sai (B)
S (B) =
|B|
Attribute reduction design of the rough set-based genetic algorithm
Rough set theory is effective when dealing with many seaport
risks and NP-hard (Nondeterministic Polynomial) problems in the
combinatorial optimization of the dataset. The rough set-based ge-
netic algorithm improves individual metrics in the evolutionary
process and avoids population diversity imbalance, by introducing a
Hamming distance when initializing the population. As a result, the
initial population may cover the entire solution space. Theorem 1
employs a constraint condition (reported in Appendix A); thus, we
assign a value equal to one to the attribute reduction set position,
according to the conditional attributes with the largest attribute
dependence. To identify the variation in the population diversity in
the evolutionary process of the algorithm, we reduce the decisional
attributes using the adaptive method of the rough set-based genetic
algorithm. We decided that the termination condition would be
reach 200 iterations, to directly analyze algorithm diversity. As the
initial population is randomly generated, the algorithm's initial
The Asian Journal of Shipping and Logistics 38 (2022) 222–233
population may not be the same, nor the evolutionary processes of
the two attribute reductions.
When using the genetic algorithm, the genetic representation
(3)
method uses a fixed n-bit binary series if the actual characteristic of
(4)
attribute reduction is considered, in line with Luo, Ji, Fu, and Tong
(2007). This study's encoding length is the number of conditional
attributes in the decision table. For instance, if the conditional at-
tribute set is {a1, a2, a3, a4, a5, …, a61}, the encoding length is 61.
Furthermore, if the reduced binary code is 10001…1, the reduction
in the decision table is {a1, a5, …, a61}. In other words, the condi-
tional attribute room C may be mapped into an individual room,
when each bit corresponds to a conditional attribute. Therefore, the
latest method in this algorithm may be processed as described in
Section 3.3.
An individual's fitness relies on two aspects: the number of at-
tributes and the ability to classify them. Both dimensions relate to
(5)
the decision-making attribute set, as follows:
|POSC (D)| |C| |B|
F (B) = p1 + p2 + p3 S¯ (B)
|C| (10)
(6)
The function consists of three parts. The first part represents the
classification ability. A reduction occurs when k = 1. The second part
shows the reduction rate. For example, the fewer attributes in the
attribute subset B indicate that B is a minimum reduction with a high
reduction rate. The third part is the weight factor, which increases
the efficiency of the algorithm. The three elements are dynamically
adjusted during algorithm evolution.
In this study, the adaptive factors are the adjustment parameters
(7) used to ensure correct reduction results with minimum reduction.
Fig. 1 explains the steps for creating a rough set-based genetic al-
gorithm. Moreover, a smaller value of p1 indicates a high possibility
that the chromosome is a reduction set, and its fitness value is
greater. As the value of p2 decreases, the conditional risk attribute set
declines, while the fitness value of the reduction increases. Finally, as
p3 increases, the attribute contained in the reduction set becomes
more relevant, and the fitness value of the reduction increases.
(8) Hence, the adaptive function of this study is defined as:
p1 = 5, p2 = 2, p3 = 1, (C , D) < (B , D)
p1 = 6, p2 = 3, p3 = 1, (C , D) = (B, D) (11)
To find the central tendency of the seaport-fulcrum supply chain
risk, we identify the 61 elements of seaport risk (Table 1). The po-
tential threat to supply chain continuity follows as the top event.
(9) Next, we employ the rough set-based genetic algorithm to reduce
the number of attributes, without losing their dependency degree.
Evaluation matrix of central tendency
The results of the rough set-based genetic algorithm belong to
the attribute reduction set. To find the central tendency in the at-
tribute reduction set, we manually use the discernibility matrix of
the rough set. Concerning the element classification in the universe,
a reduction set enables the same classification of elements of the
universe as the entire attribute set. For example, let B be a subset of
A; hence, the connecting notion between the core-set attribute and
the reduction set attribute is defined by:
Core (B) = Red (B). (12)
Eq. (12) indicates that the core attribute that becomes the center
of the tendency, is the intersection of all the reduction sets. Thus, the
core is the most relevant subset of attributes because none of its
elements may be removed without affecting the attribute classifi-
cation power.
226
M.R. Do. Bagus and S. Hanaoka The Asian Journal of Shipping and Logistics 38 (2022) 222–233

The group of core attribute sets in the new decision table is de- Table 4
ployed to calculate the degree of importance. Let S be the informa- Result of reliability test for primary dataset.
tion system table of the following tuple: S ={Corex(B), γi(B,D), Vγi}, Case Processing Summary Number of Percentage
where Corex(B) is the core attribute x in subset B, γi(B,D) is the de- responses
pendency degree of i (i=1, 2,… n; the number of algorithms running), Cases Valid 153 100%
and Vγi is the value of the dependency degree. We propose an im- Excludeda 0 0.0
portance degree, to obtain the central tendency among the core-set Total 153 100%
attributes, and the weight indicator to estimate the relative im- Reliability Statisticsb Percentage
portance using the core-set attributes as follows: Cronbach’s alpha 0.882 88.2%
Number of features 62 100%
i (B, D)
i = ,
Core x (B) U (13) a
Listwise deletion based on all variables in the procedure

i
i = . b
Cronbach’s alpha level of reliability 0.0–0.20 Less reliable
i (14) > 0.20–0.40 Rather reliable
> 0.40–0.60 Quite reliable
To estimate which seaport risk attributes have the highest im- > 0.60–0.80 Reliable
plication for the seaport-fulcrum supply chain threat factors, the > 0.80–1.00 Very reliable
threat score is calculated based on the average value of the re-
spondent evaluation for each conditional seaport risk attributes and
weight indicator in Eq. (17), as shown in the following equations:
1 u
C x¯ij = a xij;
U i = 1, j = 1 (15)

i = i × C xij . (16)

Data mining: association rule learning

Association analysis is based on the concept of rough set theory

and contains three main measures – strength, certainty, and cov-
erage – as detailed below. Strength indicates how frequently the Fig. 2. The quality of reduction.
antecedent C and consequent D both occur across the data set. In
other words, it represents the probability that conditional and de- needed to determine the weight of the parameters, and sometimes
cisional risk factors occur simultaneously as shown below. leads to a multiple attribute decision problem. This study introduces
Suppx (C , D) into the algorithm a weight indicator based on the significance de-
x (C; D) = gree of the reduction set attribute in Eq. (9). Furthermore, the re-
|U| (17)
duction attribute set is set up from 200 iterations of the rough set-
Certainty represents the conditional probability that deficiency based genetic algorithm, as shown in Fig. 2 as the quality of reduc-
itemset D occurs under the condition that deficiency itemset C oc- tion. In Fig. 2, the criteria that were met imply that the fitness
curs, which means the frequency of occurrence is found in the data function calculation in Eq. (10) fulfills the iteration set up in the
set as defined below. algorithm, while the percentage stall indicates that the calculation
Suppx (C , D) falls into the local optimal solution. Hence, this study only finds the
|C (x) D (x)|
cerx (C; D) = = central tendency among the reduction sets with 100% criteria met
|C (x)| |C (x)| (18)
and 0% stall. In line with these conditions, the evaluation matrix is
Coverage is defined as the ratio of the conditional probability of introduced to obtain the Boolean function in Table 6.
occurrence of the antecedent C and that of the consequent D to the According to Fig. 2, the set of reduction attributes is sourced from
probability of occurrence of the antecedent C as expressed below. the running algorithm as many as 50 times, and is shown in Table 5.
The corresponding matrix shown in Table 6 results in a dis-
|C (x) D (x)| Suppx (C , D)
covx (C; D) = = cernibility function and simplification with the absorption laws of
|D (x)| |D (x)| (19) Boolean algebra, as follows:

f (A) = {a15 a101a81} + {a16 a25 a54 a61} + {a73 a81a101} + {a34 a105 a101}
Results and discussion
+ {a81} + {a 43 a105} + {a63 a73 a94 a102} + {a14 a34 a81}
Pilot test of questionnaire survey + {a14 a15 a 42} + {a105} + {a18 a91 + a91a101} + {a91}, (20)

To begin the proposed model, the initial test of primary data was
deployed first to check the reliability of the dataset after collection.
There are 61 conditional seaport risk factors and a decisional factor
from 153 responses which were tested. Table 4 shows the reliability
test results. According to the Cronbach’s alpha, the dataset was very
reliable as an input for the rough set-based genetics algorithm.
where + denotes the Boolean sum, while Boolean multiplication is
omitted.
From Eq. (22), we obtain the core attribute sets or the central
tendency of the seaport-fulcrum supply chain risk reported in Table 7.
This study simply selects the core attributes from the Eq. (22) and
aggregates all of them into the top event of dimensional threats.

Multi-factor set of threat dimension evaluation Reduction performance analysis

Indexing the set is a crucial construction related to the reason- To analyze the convergence performance of the algorithm, we
ability and accuracy of the comprehensive evaluation. This step is record the alterations in the optimal individual and average fitness

227
M.R. Do. Bagus and S. Hanaoka The Asian Journal of Shipping and Logistics 38 (2022) 222–233

Table 5
Set 48 The reduction attribute set from the result of the rough set-based genetic algorithm.

a102

a91

a91
a15

a15
No. Reduction attribute sets No. Reduction attribute sets
–
–
–
–
–

–
–
a39, a73, a81 1 a15, a24, a26, a48, a81, a101 26 a14, a34, a64, a81, a82, a96
2 a42, a61, a92, a102, a104, a105 27 a34, a43, a52, a92, a94, a104
3 a13, a21, a63, a66, a91, a102 28 a26, a41, a72, a82, a93, a103
Set 46

4 a16, a25, a31, a54, a61, a103 29 a14, a15, a42, a61, a94, a96
a81

a81

a81
a42
a73

a91

a91
–

–
–
5 a34, a39, a53, a73, a81, a101 30 a11, a19, a24, a34, a35, a72, a81
6 a44, a51, a96, a103, a104 31 a15, a27, a31, a38, a71, a91
a155, a105

7 a18, a38, a39, a43, a92, a105 32 a14, a38, a61, a64, a94, a105
Set 45

8 a15, a25, a47, a52, a91, a93 33 a21, a33, a34, a51, a73, a101
a105

a105
a54

9 a13, a23, a41, a46, a62, a72, a95 34 a52, a53, a82, a91, a102, a104
–

–
–
–

–
–

–
–
10 a33, a34, a41, a43, a101, a105 35 a31, a62, a63, a72, a92, a103
11 a44, a51, a96, a103, a104 36 a26, a31, a53, a54, a82, a91
a26, a101

a43, a101

a18, a101
Set 41

12 a18, a38, a39, a43, a92, a105 37 a18, a34, a54, a91, a101
a101

13 a15, a25, a47, a52, a91, a93 38 a15, a24, a72, a81, a91, a101
a43
–

–
–
–
–

–
–
–
14 a13, a23, a41, a46, a62, a72, a95 39 a26, a27, a34, a82, a96, a101
15 a33, a34, a41, a43, a101, a105 40 a12, a15, a19, a22, a43, a96, a103
a34, a101
a34, a101

a18, a101

16 a16, a31, a39, a66, a72, a81 41 a18, a23, a26, a43, a62, a101
Set 37

a101

17 a15, a17, a25, a91, a94, a103 42 a17, a19, a32, a44, a66, a94
a54

a34

a91
a91
–
–
–

–
–

18 a14, a26, a36, a48, a66, a83, a101 43 a27, a32, a42, a63, a94, a102
19 a18, a47, a62, a63, a95, a101 44 a19, a33, a44, a55, a105
a14, a61, a94

20 a16, a31, a39, a66, a72, a81 45 a35, a44, a54, a95, a104, a105
a38, a105

21 a33, a38, a43, a95, a105 46 a26, a39, a42, a73, a81, a91
a14, a62,
a62, a94
Set 32

22 a15, a17, a25, a91, a94, a103 47 a19, a24, a38, a42, a61, a102, a103
a105

a105
a61

23 a25, a37, a64, a73, a94, a102 48 a15, a21, a22, a36, a91, a102
–

–
–

–
–

24 a14, a24, a38, a44, a94, a105 49 a14, a35, a43, a46, a96, a102, a105
25 a18, a34, a64, a71, a103, a105 50 a14, a34, a52, a95, a104
a14, a61, a94
a14, a96
Set 29

values, in the iterative process for each reduction attribute set. The
a81

a94

a42
a61

a15
–
–
–
–

–
–
–

convergence speed is stable after ten generations from running the

algorithm 50 times, indicated in Figs. 3 and 4, respectively. Because
a14, a62,
a34, a81

a14, a96
Set 26

of the elitist strategy, the convergence speed of the rough set-based

a81

a34
a81

a62

a34

a81
–

–
–

genetic algorithm accelerates the evolution mechanism. This

strategy copies the individuals with the highest fitness value directly
a62, a94
Set 23

into the next generation, without manipulation in the reduction set.

a102
a25

a62
a94
a73

a73

As an impact, the algorithm converges with the global optimal so-

–

–
–
–

–
–
–

lution.
a33, a43, a105

a38, a105

a95, a105
Set 21

Evaluation of attribute ranking and potential threats

a43
–
–
–

–
–
–

–
–

Each attribute in Table 7 indicates the central tendency of the

a39, a81
a16, a31

Indonesian seaport-fulcrum supply chain risk. These central ten-

Set 16

dencies are classified on the basis of dimensional threats. For man-

a81

a81

a81
–

–
–

–
–
–
–
–

agement analysis purposes, we employ Eqs. (13)–(16) to calculate

the risk score for each dimensional threat. In addition, the scores of
a33, a43, a105

the core attributes reflect the feasibility of the attributes. The higher
a34, a101

a34, a101
a43, a101
Set 10

the attribute score, the lower tendency for the seaport-fulcrum

a101

a105

a105
a34

supply chain. The ranking of the potential threats to supply chain

–

–
–

disruption is depicted in Fig. 5.

a39, a73, a81

We consider the issue of supply chain coordination related to

a81, a101

a34, a101

a34, a101
a39, a81

a34, a81

disruptive events under the condition of satisfying self-organized

Set 5

a101

criticality. It is worth mentioning that the probabilities of all po-

a73
–

–
–

tential threats cannot be derived. However, the seaport risk score

The discernibility matrix of reduction attribute sets.

helps understand the feasibility level of the potential threats to

a16, a31
Set 4

supply chain disruptive events. In Fig. 5, σi indicates the number of

a25

a54

a54
a61
a61
–

–
–

–
–

scores from each dimensional threat, and σi(Total) indicates the re-
sult obtained by subtracting σi from the total of σi divided by two. As
a81, a101

a26, a101

the threat feasibility declines due to the effect of monotonicity in Va,

Set 1

a101

a101
a81

a81
a81

a81
a15

σi(Total) rises. Hence, the trendline of the power-law helps us

–

–
–

identify the probability of potential threats. This convex function

Reduction attribute sets

implies that they have a negative slope (dimensional threats). Hence,

we typically assume that a higher score is clear, and a lower score is
unclear. The threats from the planning process (A1), infrastructure
(A2), distribution process (A4), nuclear enterprise financial (A6), lo-
cation (A8), and security (A9) are clear in terms of risk probability,
while the threats from the seaport service process (A3), relationship
23
26
29
32

45
46
48
37
21

41
10
16
Table 6

1
4
5

process (A5), monetary (A7), and environmental (A10) are beyond

Set
Set
Set
Set
Set
Set
Set
Set
Set
Set
Set
Set
Set
Set
Set

stakeholder comprehension. Hence, the ten-dimensional threats

228
M.R. Do. Bagus and S. Hanaoka The Asian Journal of Shipping and Logistics 38 (2022) 222–233

Table 7
The score of the seaport-fulcrum supply chain risk.

No. Dimensional threats Attributes Importance degree ( ij ) Weight ( ij ) Evaluation average value (C xij ) Score ( ij )

1 A1 a14 0.02335 0.04774 2.57895 0.1231

2 a15 0.02069 0.04230 2.60526 0.1102
3 a16 0.01495 0.03056 2.41447 0.0738
4 a18 0.01400 0.02863 2.75658 0.0789
5 A2 a25 0.07299 0.14923 2.58882 0.3860
6 a26 0.01851 0.03784 3.35526 0.1270
7 A3 a31 0.01843 0.03769 3.26974 0.1232
8 a33 0.03694 0.07553 3.31250 0.2502
9 a34 0.01388 0.02838 2.76316 0.0784
10 a38 0.01600 0.03271 2.76974 0.0906
11 a39 0.02453 0.05015 2.67763 0.1343
12 A4 a42 0.01466 0.02997 2.61184 0.0783
13 a43 0.02271 0.04643 2.61184 0.1213
14 A5 a54 0.09177 0.18764 2.68684 0.5029
15 A6 a61 0.01495 0.03057 2.52632 0.0772
16 a64 0.02024 0.04138 2.58553 0.1070
17 A7 a73 0.03519 0.07195 2.55592 0.1842
18 A8 a81 0.02182 0.04462 2.49342 0.1113
19 A9 a91 0.02151 0.04398 2.67105 0.1175
20 a94 0.02126 0.04346 2.50000 0.1087
21 a95 0.04277 0.08744 2.58553 0.2261
22 A10 a101 0.02094 0.04282 2.56579 0.1099
23 a102 0.03244 0.06634 3.21053 0.2130
24 a105 0.02028 0.04147 3.00658 0.1247

Fig. 3. Optimal adaptation process.

mean that the seaport risk poses potential threats in various di-
mensions to disrupt supply chain continuity in Fig. 6.
The lack of storage risk planning is ranked highest among 24
central tendencies of seaport-fulcrum supply chain risk disruption in
Fig. 5. For example, Siswanto, Kurniawati, Latiffianti, Rusdiansyah,
and Sarker (2018) highlighted the problem of storage planning par-
ticularly in the fertilizer product supply chain, clearly implicating an
inventory routing problem. Initially, the fertilizer company chartered
a vessel for some time horizon under the capacity constraint. The
demurrage in the loading port implicated a decline of the ship’s
utility and increased the ship’s operational cost. The former clearly
affected congestion in the waterways, whereas the latter affected
low growth for the factory.
This main risk of the dimensional planning process threat is re-
lated to seaport infrastructure and management (Loh & Thai, 2015).
Despite the significant effort required to develop seaport infra-
structure, the result will improve its performance with high space
utilization. The utilization rate of these areas should be tracked to
ensure that new area plans are carried out on time. Furthermore, the
risk of a port confronting a dimensional threat of the planning
process threat increases if it is either a regional gateway owing to
less timely customs clearance, or a transshipment hub port due to
less cost efficiency in feeder links and short travel times to the other
hub ports. In this instance, the hub port may encounter disturbed
working schedules if vessels' arrivals are delayed owing to inclement
weather, inventory routing, and security problems at earlier ports. A

229
M.R. Do. Bagus and S. Hanaoka
Fig. 4. Alteration process of the average fitness.
Fig. 5. Th ranking of the seaport-fulcrum supply chain risks.
Fig. 6. The potential threats to the seaport-fulcrum supply chain.
hub port also has a busy waterway since it feeds a network of ves-
sels, some of which may choose to obtain supplementary services,
such as water supply and bunkering, while docked in seaport waters,
while others might experience vessel bunching.
On the other hand, low punctuality of delivery goods, which was
ranked second in the central tendency of the supply chain, results
from various risk factors that have significant implications in the
supply chain distribution process. Dewi and Purnamasari (2021)
revealed that low punctuality of delivery goods is caused by in-
efficient maritime security inspections at sea as part of seaport
customs clearance operations. Consequently, the shipping industry
230
The Asian Journal of Shipping and Logistics 38 (2022) 222–233
suffers operational delays and increased exposure to liabilities under
their contract of carriage. Furthermore, stoppages for maritime se-
curity checks at sea generate delays, which raise shipping expenses.
If a ship is delayed due to a marine security inspection, the delayed
ship will need to speed up to meet the same arrival schedule at its
next port and will burn more fuel than expected leading to green-
house emissions. If its arrival schedule at the destination port
changes as a result of the delay, it may have to pay to reschedule
services, including pilotage, class inspections, and planned main-
tenance.
Generally, the seaport-service is a main duty in seaport opera-
tions. This study found that the port capacity shortage, waterways
congestion, and hinterland transfer congestion are among the top
ten central tendencies that pose threats to the seaport service pro-
cess. Together with the monetary and location threats, the shipping
industry may consider this issue as a port-choice problem.
Indonesia, as an archipelago country, faces a difficult challenge in
achieving maritime logistics system efficiency. The higher logistics
costs account for around 20 – 30% of Indonesia’s GDP because of the
archipelagic characteristics particularly in eastern Indonesia (Amin
et al., 2021). This research shows that a market imbalance in terms of
container and cargo throughput exists in the Indonesian maritime
supply chain. The availability of the manufacturing industry and
medium enterprises also contributes to the market imparity.
Meanwhile, irregular ship schedules, high sea transportation costs,
and lack of transportation modes are some issues faced by the
maritime supply chain in eastern Indonesia as a result of market
imbalance.
Seaport location is also important in the global supply chain
since many shipping lines try to maintain a small number of ports of
call with maximum market coverage, especially if they are as close
to the growth market as possible (Kavirathna et al., 2018). Short
sailing time to other hub ports is considered a central tendency of
location threat in the Indonesian seaport-fulcrum supply chain op-
eration. Such a phenomenon, known as Short Sea Shipping (SSS),
provides cost competitiveness in the modal shift from road haulage
for the seaport-users. However, this also brings a negative ripple
effect to the shipping industry and seaport management. The former
might draw excessive competition in the shipping market to the
seaport-fulcrum supply chain society. Excessive competition, for
example, may result in the extinction of shipping lines with weak
M.R. Do. Bagus and S. Hanaoka
Fig. 7. Decision class 1 (the highest implication of risk to disrupt supply chain continuity).
financial conditions, while surviving shipping businesses will use
their monopolistic status to hike rates. This will soon have an ad-
verse effect on the shipper and society. Meanwhile, the latter in-
duces congestion within the terminal since the number of container
throughputs in the stacking yard increases. Moreover, the more
frequent presence of vessels in hub ports due to short sailing times
increases greenhouse emissions.
The aforementioned potential threats directly or indirectly im-
pacted the financial condition of the seaport-fulcrum supply chain
members. Revenue and growth development are also central ten-
dencies in particular cases. Both are primary tendencies of financial
risk precautions for each member of the seaport-fulcrum supply
chain threat. Hence, decreasing financial loss, preventing capital
chain breakdown, and avoiding nuclear-enterprise financial risk are
compulsory attempts to establish risk retention and risk prevention
along the supply chain entities (Jiang et al., 2018).
Relationship between the decision class and risk level
In the analysis, five decision classes are based on the rough set
theory approach employed to understand risk factor level. However,
only decision class 1 (highest implication to disrupt supply chain
continuity) was analyzed with the five threat risk levels. The risk
level according to Fig. 5 is then obtained based on the score assigned
by the seaport supply chains’ stakeholder. The connections between
decision class 1 and risk level according to Eqs. (17) – (18) are de-
picted in Fig. 7.
The frequency among 152 seaport-fulcrum supply chain stake-
holders on seaport risk levels is depicted in Fig. 7. Moreover, the
central tendency of seaport risks is segmented according to their risk
level and shown by three indicator measurements. At the highest
231
The Asian Journal of Shipping and Logistics 38 (2022) 222–233
risk level, a level of over 40% in the strength indicator and coverage
indicator, such as less timeliness of port customs clearance, low
growth development, less efficient costs in feeder links, less moti-
vation of member interest distribution mechanisms, low revenue,
shortage of IT and advanced technology, and lack of ship risk plan-
ning, is likely to signify the highest disruption in this supply chain
context.
Conclusion
In this study, the seaport-fulcrum supply chain risk was analyzed
using a rough-set-based genetic algorithm and data from a ques-
tionnaire survey. The proposed method can support study of the
combination problem among seaport risk factors as well as between
many seaport risk factors and supply chain threats from the per-
spectives of seaport-manager, seaport-operator, and seaport-user
evaluations, which can provide insights into risk management at the
seaport. The score aids stakeholders in reacting correctly based on
the seaport risks and threats.
According to the study findings, 24 seaport risks in Indonesia
pose potential threats to supply chain continuity. In terms of feasi-
bility, the seaport service process (A3), relationship process (A5),
monetary (A7), and environmental (A10) are beyond stakeholder
comprehension. The shortage of port capacity, congestion in water-
ways, and congestion in hinterland transport are among top-ten
priorities in the central tendency of the Indonesia seaport-fulcrum
supply chain risk disruption and induced threat to the seaport ser-
vice process. Less motivation of member interest distribution me-
chanism also poses a threat to the relationship process among
supply chain entities, whereas less efficient costs in the feeder link is
considered a monetary threat. The occurrence of pandemics/
M.R. Do. Bagus and S. Hanaoka
epidemics (COVID-19), which is another top-ten priority of the
central tendency list, is considered an environmental threat.
Several aspects of the Indonesian seaport-fulcrum supply chain
operation are in danger of interruption. First, seaport risk factors
from various dimensional threat categories, such as planning process
(A1), infrastructure (A2), distribution process (A4), nuclear enterprise
financial (A6), location (A8), and security (A9), influenced other ca-
tegories previously, making threats more vague and unpredictable.
For example, in a specific time frame, there was no planning process
or dependable planning procedure. As a result, the unpredictability
of the ports service process-related hazard increased. Second, the
lack of cooperation among supply chain entities in carrying out their
tasks, caused by the lack of incentive of the member sharing interest
distribution mechanism, has increased. Furthermore, the container
throughput imbalance caused by inefficiency of costs in feeder
connections, notably in Indonesia's eastern region, becomes a
monetary danger that should be addressed. Finally, the occurrence of
COVID-19 has considerably disrupted the supply chain operation at
the seaport.
Ultimately, a supply chain emergency management system, as
well as a contingency plan and emergency protection mechanism for
force majeure events such as seaport risk, should be developed to
decrease risk and prevent unanticipated occurrences. The seaport
management is also strongly urged to work closely with its stake-
holders and the government to ensure that future plans keep up
with new trends and are related to consumer requests, as well as
national trade and investment goals. Relevant information, such as
client vessel size orders, shipping market assessments, and chal-
lenges encountered by seaport users, should be communicated with
the seaport-operator through the seaport-manager, since these have
an influence on the adequacy of future plans. In addition, a supply
chain management mechanism for the seaport enterprise core
should be established; strategic cooperators should be engaged to
improve the ability of upstream and downstream supply chain en-
terprises to deal with unexpected incidents and prevent supply
chain breakdown as a result of these factors.
On the other hand, insufficiencies still exist that should be ex-
amined in future research. For instance, the importance rating of
each potential threat dimension is not clearly explained. Moreover,
the risk of seaport disruption to a supply chain’s operation is de-
termined by many factors when operating in a complex socio-
technical environment. Sensitivity analysis from a pair-wise function
is thus needed to determine to what extent the seaport-focal supply
chain risk model can be analyzed.
Funding
This study was funded by the Indonesia Endowment Fund for
Education (LPDP), Ministry of Finance, Republic of Indonesia
(Kementerian Keuangan Republik Indonesia) through a scholarship
scheme (Grant No. S-1975/LPDP.4/2022).
Declarations of Interest
The authors declare that they have no known competing fi-
nancial interests or personal relationships that could have influ-
enced the work reported in this paper.
Acknowledgments
This study was supported by the Indonesia Endowment Fund for
Education (LPDP), Ministry of Finance, Republic of Indonesia
(Kementerian Keuangan Republik Indonesia) and would not have
been possible without support from the Indonesian Seaport
Organizations.
232
The Asian Journal of Shipping and Logistics 38 (2022) 222–233
Disclosure statement
The authors reported no potential conflict of interest.
Appendix A. The Theorem and Proof of Eq. (8)
Theorem 1. Suppose ai ∈ β, where β denotes the attribute reduction
set. Then, Max (B (D)), which follows the condition attribute ai
belonging to the condition set of attributes B (where B ⊆ A), is D-
superfluous if it has no effect on the lower approximation of D.
Otherwise, attribute ai is D-indispensable in A.
Proof 1. As defined in Section 3.3, for a ∈ C, the
SB (D) = C (D) B ai (D) ; if k= 1, D depends entirely on C, the SB (D)
becomes larger, and the B ai (D) decreases, as mathematically
defined by Eq. (3). Furthermore, the smaller |POS(B ai) (D)|, the
greater the dependency of decision attribute D on attribute a, and
|POS(C a ) (D)|
i
the larger (C , D), as mathematically shown in SC (D) = |U|
,
where |U| is a fixed value. Owing to the addition of attributes in the
order of significance, a reduction may be obtained, so that Max (IB
(D)) ∀ ai ∈ β. The set of D-indispensable attributes in A is called the D-
core of A. However, the minimal subsets of conditional attributes
that discern all equivalence classes of the relation Ind(D), discernible
by the entire set of attributes, are called D-reducts. In other words, if
each conditional attribute in the decision table is independent of D,
then the conditional attribute set C is independent of D; otherwise, C
is dependent on D.
Appendix B. The Proof of Eq. (9)
Proof 2. Let the conditional attribute set be C and B be a subset of C.
As shown in Eq. (8), for ai element B, each attribute in B is
independent of D, and S (B) becomes larger. Otherwise, B ai (D)
becomes smaller than in Eq. (8). Thus, B is more likely to be reduced.
Specifically, when POSB(D) = POSC(D), B is the reduction in C
according to the definition in Eqs. (3) and (4).
References
Amin, C., Mulyati, H., Anggraini, E., & Kusumastanto, T. (2021). Impact of maritime
logistics on archipelagic economic development in eastern Indonesia. Asian
Journal of Shipping and Logistics, 37(2), 157–164. https://doi.org/10.1016/j.ajsl.2021.
01.004
Aqlan, F., & Lam, S. S. (2015). A fuzzy-based integrated framework for supply chain risk
assessment. International Journal of Production Economics, 161, 54–63. https://doi.
org/10.1016/j.ijpe.2014.11.013
Bichou, K., & Gray, R. (2005). A critical review of conventional terminology for clas-
sifying seaports. Transportation Research Part A, 39(1), 75–92. https://doi.org/10.
1016/j.tra.2004.11.003
Bogataj, D., & Bogataj, M. (2007). Measuring the supply chain risk and vulnerability in
frequency space. International Journal of Production Economics, 108(1–2), 291–301.
https://doi.org/10.1016/j.ijpe.2006.12.017
Dewi, Y. K., & Purnamasari, D. (2021). Costs of maritime security inspection to mer-
chant ship operations – the Indonesian shipowners’ perspective. Australian
Journal of Maritime and Ocean Affairs, 1–16. https://doi.org/10.1080/18366503.
2021.1962059
Esteban, M., Takagi, H., Nicholls, R. J., Fatma, D., Pratama, M. B., Kurobe, S., & Avelino, E.
(2020). Adapting ports to sea-level rise: empirical lessons based on land sub-
sidence in Indonesia and Japan. Maritime Policy and Management, 47(7), 937–952.
https://doi.org/10.1080/03088839.2019.1634845
Hall, P. V. (2007). Seaports, urban sustainability, and paradigm shift. Journal of Urban
Technology, 14(2), 87–101. https://doi.org/10.1080/10630730701531757
Jiang, B., Li, J., & Shen, S. (2018). Supply Chain Risk Assessment and Control of Port
Enterprises: Qingdao port as case study. Asian Journal of Shipping and Logistics,
34(3), 198–208. https://doi.org/10.1016/j.ajsl.2018.09.003
John, A., Paraskevadakis, D., Bury, A., Yang, Z., Riahi, R., & Wang, J. (2014). An integrated
fuzzy risk assessment for seaport operations. Safety Science, 68, 180–194. https://
doi.org/10.1016/j.ssci.2014.04.001
Kavirathna, C., Kawasaki, T., Hanaoka, S., & Matsuda, T. (2018). Transshipment hub port
selection criteria by shipping lines: the case of hub ports around the bay of
Bengal. Journal of Shipping and Trade, 3(1), https://doi.org/10.1186/s41072-018-
0030-5
M.R. Do. Bagus and S. Hanaoka
Perhubungan, Kementerian (2020). Transportation statistics books from 2010 until
2020. Retrieved from https://dephub.go.id/public/images/uploads/posts/statistik-per-
hubungan-jilid-ii-2020.pdf.
Klibi, W., & Martel, A. (2012). Scenario-based Supply Chain Network risk modeling.
European Journal of Operational Research, 223(3), 644–658. https://doi.org/10.1016/
j.ejor.2012.06.027
Komarudin, K., Reza, M., Moeis, A. O., & Rahmawan, A. (2017). Enhancing pendulum
Nusantara model in Indonesian maritime logistics network. Journal of Traffic and
Logistics Engineering. https://doi.org/10.18178/jtle.5.1.35-39
Loh, H. S., & Thai, V. V. (2015). Management of disruptions by seaports: preliminary
findings. Asia Pacific Journal of Marketing and Logistics, 27(1), 146–162. https://doi.
org/10.1108/APJML-04-2014-0053
Loh, H. S., Zhou, Q., Thai, V. V., Wong, Y. D., & Yuen, K. F. (2017). Fuzzy comprehensive
evaluation of port-centric supply chain disruption threats. Ocean & Coastal
Management, 148, 53–62. https://doi.org/10.1016/j.ocecoaman.2017.07.017
Luo, K., Ji, H., Fu, P., & Tong, X., 2007, A new method based on genetic algorithm for
reduction of attribution under incomplete decision-making table. In Third
International Conference on Natural Computation (ICNC 2007). IEEE Publications.
Mokhtari, K., Ren, J., Roberts, C., & Wang, J. (2012). Decision support framework for
risk management on seaports and terminals using fuzzy set theory and evidential
reasoning approach. Expert Systems with Applications, 39(5), 5087–5103. https://
doi.org/10.1016/j.eswa.2011.11.030
Morris, L. L. (2020). Stakeholder collaboration as a pathway to climate adaptation at
coastal ports. Maritime Policy and Management, 47(7), 953–967. https://doi.org/10.
1080/03088839.2020.1729435
Notteboom, T., Pallis, T., & Rodrigue, J.-P. (2021). Disruptions and resilience in global
container shipping and ports: the COVID-19 pandemic versus the 2008–2009
233
The Asian Journal of Shipping and Logistics 38 (2022) 222–233
financial crisis. Maritime Economics and Logistics, 23(2), 179–210. https://doi.org/
10.1057/s41278-020-00180-5
Radivojević, G., & Gajović, V. (2014). Supply chain risk modeling by AHP and Fuzzy
AHP methods. Journal of Risk Research, 17(3), 337–352. https://doi.org/10.1080/
13669877.2013.808689
Rumaji, & Adiliya, A (2019). Port maritime connectivity in south-east Indonesia: A new
strategic positioning for transhipment port of tenau kupang. Asian Journal of
Shipping and Logistics, 35(4), 172–180. https://doi.org/10.1016/j.ajsl.2019.12.004
Singh, R. (2017). Hazards and threats to a pipeline system. Pipeline Integrity Handbook.
Amsterdam: Elsevier,35–88.
Siswanto, N., Kurniawati, U., Latiffianti, E., Rusdiansyah, A., & Sarker, R. (2018). A
Simulation study of sea transport based fertilizer product considering disruptive
supply and congestion problems. Asian Journal of Shipping and Logistics, 34(4),
269–278. https://doi.org/10.1016/j.ajsl.2018.08.001
Sudha, M. (2017). Evolutionary and neural computing based decision support system
for disease diagnosis from clinical data sets in medical practice. Journal of Medical
Systems, 41(11), 178. https://doi.org/10.1007/s10916-017-0823-3
Weng, J., Liao, S., Wu, B., & Yang, D. (2020). Exploring effects of ship traffic char-
acteristics and environmental conditions on ship collision frequency. Maritime
Policy and Management, 47(4), 523–543. https://doi.org/10.1080/03088839.2020.
1721584
Wu, C., Yue, Y., Li, M., & Adjei, O. (2004). The rough set theory and applications.
Engineering Computations, 21(5), 488–511. https://doi.org/10.1108/
02644400410545092
Zaman, M. B., Vanany, I., & Awaluddin, K. D. (2015). Connectivity analysis of port in
eastern Indonesia. Procedia Earth and Planetary Science, 14, 118–127. https://doi.
org/10.1016/j.proeps.2015.07.092