Content uploaded by Yongrui Guo
Author content
All content in this area was uploaded by Yongrui Guo on Apr 26, 2022
Content may be subject to copyright.
This article was downloaded by: [Stephen F Austin State University]
On: 05 August 2015, At: 04:23
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5
Howick Place, London, SW1P 1WG
Click for updates
Asia Pacific Journal of Tourism Research
Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/rapt20
Modeling the Fluctuation Patterns of Monthly
Inbound Tourist Flows to China: A Complex
Network Approach
Yongrui Guoa, Jie Zhanga, Yang Yangb & Honglei Zhanga
a Department of Land Resources and Tourism Sciences, Nanjing University,
Nanjing, Jiangsu, People's Republic of China
b School of Tourism and Hospitality Management, Temple University,
Philadelphia, PA, USA
Published online: 26 Aug 2014.
To cite this article: Yongrui Guo, Jie Zhang, Yang Yang & Honglei Zhang (2015) Modeling the Fluctuation
Patterns of Monthly Inbound Tourist Flows to China: A Complex Network Approach, Asia Pacific Journal of
Tourism Research, 20:8, 942-953, DOI: 10.1080/10941665.2014.948024
To link to this article: http://dx.doi.org/10.1080/10941665.2014.948024
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed in this publication
are the opinions and views of the authors, and are not the views of or endorsed by Taylor &
Francis. The accuracy of the Content should not be relied upon and should be independently
verified with primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use
can be found at http://www.tandfonline.com/page/terms-and-conditions
Modeling the Fluctuation Patterns of Monthly
Inbound Tourist Flows to China: A Complex
Network Approach
Yongrui Guo
1
, Jie Zhang
1
, Yang Yang
2
and Honglei Zhang
1
∗
1
Department of Land Resources and Tourism Sciences, Nanjing University, Nanjing, Jiangsu,
People’s Republic of China
2
School of Tourism and Hospitality Management, Temple University, Philadelphia, PA, USA
A thorough understanding of the fluctuations of tourist flows provides useful insights con-
cerning the nature of tourist demand. This study aims to investigate the fluctuation pat-
terns and dynamics of inbound tourist flows to China using a complex network
approach. Several measures, such as the network topological parameters of degree and
degree distribution, betweenness centrality, and shortest path length, are utilized to dis-
cover important fluctuation patterns and the transition distance. Based on the empirical
results, six important fluctuation patterns of inbound tourist flows to China are recog-
nized. These fluctuation patterns are important intermediaries in the process of transform-
ation of the fluctuation patterns and can be viewed as a prelude to changes in the inbound
tourist flows. The value of 3.38 found for the average transition distance suggests that the
transformation occurs approximately every three to four quarters. These findings are
useful for understanding the inherent laws and transformations governing fluctuations
in tourist flows.
Key words: inbound tourist, tourist flows, fluctuation patterns, complex network, China
Introduction
Since China adopted the “open-door” policy
in 1978, tourism, especially inbound tourism,
has developed rapidly (Yang & Wong,
2013). In 1986, tourism, as an economic
industry, was incorporated for the first time
into the five-year state plan for the National
Economy and Social Development; in 1992,
tourism was included as one of the key indus-
tries in the tertiary sector; in 1998, tourism
was selected as a new growth pole of the
national economy; and in 2009, the State
Council of the People’s Republic of China
Asia Pacific Journal of Tourism Research, 2015
Vol. 20, No. 8, 942–953, http://dx.doi.org/10.1080/10941665.2014.948024
∗Corresponding author. Email: zhanghonglei@nju.edu.cn
#2014 Asia Pacific Tourism Association
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
issued a directive to upgrade tourism to a stra-
tegic pillar industry in the national economy.
Inbound tourist flows to China increased from
27.46 million in 1990 to 132.40 million in
2012, representing an annual growth rate of
more than 7.4% (CNTA, 2013). Inbound
tourism, with its important role in securing
foreign exchange earnings, is of great impor-
tance to the economy of China (Yang &
Wong, 2013). Inbound tourism receipts
increased from US$2,217.58 million in 1990
to US$50,028.00 million in 2012 (CNTA,
2013). In view of the rapid increase in
inbound tourist flows and receipts over the
past few decades, a comprehensive examination
of fluctuations of inbound tourist flows is of
importance to both tourism business prac-
titioners and tourism policy-makers. In this
article, we introduce a complex network
approach to model the fluctuation patterns of
monthly inbound tourist flows to China.
Temporal fluctuations of tourist flows trig-
gered by seasonality and business cycles are
one of the most significant characteristics of
tourism (Assaf, Barros, & Gil-Alana, 2011;
Cuccia & Rizzo, 2011; Fourie & Santana-
Gallego, 2011;Nadal,Font,&Rossello,
2004; Song & Li, 2008;Vergori,2012). Over
the past three decades, many studies on
tourism demand analysis and forecasting have
contributed significantly to our understanding
of the temporal fluctuations of tourism
demand (Song, Li, Witt, & Athanasopoulos,
2011). However, although the importance of
temporal fluctuations of tourist flows has been
broadly recognized, it has also been acknowl-
edged that this phenomenon is not well under-
stood (De Cantis, Ferrante, & Vaccina, 2011;
Higham & Hinch, 2002;Jang,2004;Lim&
McAleer, 2001). Most existing studies of
tourism demand have involved a forecasting
perspective, and fewer studies have focused
on the fluctuation patterns of tourist flows
(Chan, Lim, & McAleer, 2005;Cho,2009).
An in-depth understanding of the fluctuations
of tourist flows is the basis for accurate predic-
tions of future trends in tourism demand. It is
important to scrutinize changes in both the
patterns and the amplitude of fluctuations (De
Cantis et al., 2011). The aim of this study is
to examine the fluctuation patterns, amplitude,
and dynamics of monthly inbound tourist flows
to China from January 1990 to December 2012
using a complex network approach.
The paper is organized as follows. After this
introductory section, we briefly outline the
recent literature relating to the study of
tourism demand. The subsequent section
describes our methodology and data source.
Using a complex network approach, we
analyze the fluctuation patterns, amplitude,
and dynamics of inbound tourist flows to
China. The results are presented in the next
section, and some concluding remarks are
offered in the final section.
Literature Review
Tourism demand forecasting is one of the most
important task for planning development and
operational management in the tourism indus-
try. Exploring the fluctuation patterns and
accurately forecasting the future tourist flows
are essential to determine successful invest-
ments for both the public and the private
sectors (Chang & Liao, 2010). The information
from these investigations and forecasts plays a
highly important role in formulating national
tourism development policy and strategic plan-
ning, optimizing allocation of tourism market
resources, and conducting decision-making for
tourism businesses (Tao & Ni, 2010).
Measuring and analyzing fluctuation is an
important aspect of the study of tourism
demand (Chu, Yeh, & Chang, 2014;Turner
Modeling the Fluctuations of Tourist Flows 943
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
&Witt,2001). Gil-Alana (2005) examined
monthly international tourist flows in the USA
by assuming seasonal univariate long-memory
processes and suggested that the total number
of arrivals implied long memory and mean
reverting behavior. Shareef and McAleer
(2005) analyzed the conditional volatility of
tourist flows to small island tourism economies
and found that the logarithm of monthly inter-
national tourist arrivals was stationary. Yan
and Wall (2003) identified the structure, charac-
teristics, and intensity of fluctuations in the
number of international visitors to China from
1980 to 1998 and showed that the overall
trend was strong annual growth, influenced by
a cyclical fluctuation. Several studies have
applied sophisticated methods for forecasting
tourism demand, such as the space– time
cluster approach (Gursoy, Parroco, & Scuderi,
2013), multivariate exponential smoothing
(Athanasopoulos & de Silva, 2012), evolution-
ary fuzzy systems (Hadavandi, Ghanbari, Sha-
hanaghi, & Abbasian-Naghneh, 2011), the
autoregressive integrated moving average
model (Coshall, 2006; Gustavsson & Nord-
strom, 2001), dynamic almost ideal demand
system approach (Kuo, Liu, & Chen, 2014),
the seasonal autoregressive integrated moving
average model (Goh & Law, 2002), the vector
autoregressive model (Song & Witt, 2006), the
autoregressive distributed lag model (Song,
Lin, Zhang, & Gao, 2010), the Lagrange multi-
plier unit root tests (Lean & Smyth, 2009), and
time-varying parameter error correction model
(Li, Wong, Song, & Witt, 2006). Several
studies have also compared the forecasting
accuracy of various models and approaches
(Cho, 2003; Kim & Moosa, 2001; Wong,
Song, Witt, & Wu, 2007). Two extensive
reviews in this area are available from Song
and Li (2008) and Li, Song, and Witt (2005).
Identification of the fluctuation character-
istics of time series data is of crucial impor-
tance in a wide variety of fields. Many
methods, such as the Lyapunov exponent,
the autoregressive conditional heteroscedasti-
city model, and the stochastic volatility
model, have been used to analyze these charac-
teristics. These traditional methods focus
primarily on the overall features of the time
series but cannot provide nuanced information
on the determinations of the system properties
(Yang, Pan, & Song, 2014;Zhang,Zhou,
Jiang, & Wang, 2010). The study of time
series data using a complex network approach
has attracted great interests among scholars.
Time series can be mapped as a complex
network using various methods, such as the
visibility graph algorithm (Lacasa, Luque, Bal-
lesteros, Luque, & Nuno, 2008) and the
coarse-graining process (Li & Wang, 2007).
Through the application of the complex
network approach, the temporal dynamics of
time series data is encoded into the topology
of the corresponding networks. According to
the statistical properties of the network, the
determinations of different fluctuation patterns
can be identified (Zhang et al., 2010). In finan-
cial time series, for example, Bonanno et al.
(2004)showed that a network can be obtained
by a correlation-based filtering procedure and
that meaningful economic information can be
extracted from noise-dressed correlation
matrices. Fluctuations or temporal imbalances
in tourist flows create an evolving complex
dynamic system. In nature and society, many
complex dynamic systems can be represented
as complex networks (Li & Wang, 2007).
Complex network approach offers a promising
new method to the analysis of tourist flow time
series data.
In this paper, we develop a weighted
network of monthly inbound tourist flows to
China from January 1990 to December
2012. The network can translate inbound
tourist flows to various characteristics in its
944 Yongrui Guo et al.
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
network topological structure. Every node
in the network corresponds to a distinct
fluctuation pattern and has a special role in
shaping the dynamics of inbound tourist
flows. We introduce several effective par-
ameters for detecting important topological
nodes of the network of inbound tourist
flows. From these nodes, we can obtain signifi-
cant fluctuation patterns of inbound tourist
flows to China. We focus on the fluctuations
and correlations of changes in inbound
tourist flows. The statistical properties of fluc-
tuations in inbound tourist flows are impor-
tant for understanding and modeling the
complex dynamics of inbound tourist flows.
Data Source and Methodology
Data Source
The monthly time series data for inbound
tourist flows to China used in the present
study were obtained from The yearbook of
China tourism statistics (1991–2013). The
series contains 276 data points from January
1990 to December 2012 (Figure 1). The term
“inbound tourists” to China refers to foreign
tourists and tourists from Hong Kong,
Macau, and Taiwan. The time series of
inbound tourist flows to China was rep-
resented as T(t)(t¼1, 2, 3, ...,N,N¼276).
Coarse-Graining Preprocess
The simplest possible method for transforming
a time series into a complex network represen-
tation is to coarse grain its range into a suitable
set of classes and to consider the transition
probabilities between these classes in terms
of a weighted network. The coarse-graining
process is an effective method for analyzing
the complexity of time series data. After a
time series interval has been divided into
homogeneous partitions, the interval can be
averaged into limited subintervals (Li &
Wang, 2007). By giving each subinterval a
symbol, the time series data can be represented
as a discrete symbolic sequence; studying the
time series is then equivalent to studying
the symbolic sequence. The coarse-graining
process maintains the fluctuation trajectory
regardless of the time series data; therefore,
Figure 1 Monthly Inbound Tourist Flows to China from January 1990 to December 2012.
Modeling the Fluctuations of Tourist Flows 945
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
the process aids in studying the complexity of
the time series data (An, Gao, Fang, Huang, &
Ding, 2014). The time series of inbound
tourist flows to China can be transformed
into a more understandable and limited sym-
bolic sequence using the coarse-graining
process. In this symbolic sequence, each
symbol denotes a distinct fluctuation pattern.
In the coarse-graining process, symbolic cat-
egories should be limited, and each symbol rep-
resents a distinct meta-pattern of fluctuation of
inbound tourist flows. The meta-pattern refers
to the distinct symbols in a symbolic sequence
(Li & Wang, 2006a). For inbound tourist
flows to China T(t), the fluctuation is
DT=Tt−Tt−1. The average monthly change
in inbound tourist flows to China from January
1990 to December 2012 is 487,690. We then
define four distinct symbols as follows:
Si=
R(DT≥487690),
r(0,DT,487690),
d(−487690 ,DT,0),
D(DT≤−487690),
⎧
⎪
⎪
⎨
⎪
⎪
⎩
(1)
where Ris a sharp-increase meta-pattern, ris a
small-increase meta-pattern, dis a small-
decrease meta-pattern, and Dis a sharp-
decrease meta-pattern. Here, R,r,d,andDare
measurements of the magnitude of the increase
and decrease of the fluctuations in inbound
tourist flows to China. Therefore, the time
series data of inbound tourist flows to China
are transformed into a symbolic sequence:
S={S1S2S3···},Si[(R,r,d,D).(2)
Network Construction
We can obtain various string combinations
using the distinctive symbols (R,r,d,D).
Each string of symbols denotes a distinct fluc-
tuation pattern of inbound tourist flows to
China. We define an n-string as a string
composed of nsymbols. For a given n, there
are a total of 4
n
different n-strings. In coarse-
graining processing, redundant information
increases as the number of symbol strings
increases (Li & Wang, 2006b). For this
reason, we defined three months (a quarter) as
a fluctuation pattern of inbound tourist flows
to China. The fluctuation patterns of inbound
tourist flows to China were 3-symbol strings
composed of R,r,d,andD. Therefore, the
fluctuation patterns and dynamics of inbound
tourist flows to China were investigated for
each six-month period. Because n¼3, 4
3
¼
64 3-strings, that is, (RRR,RRr,RRd,RRD,
RrR,RdR,RDR,... ), are theoretically poss-
ible. However, only 28 types of 3-strings actu-
ally appear. In symbol sequences of inbound
tourist flows to China, the 3-strings can rep-
resent different fluctuation patterns.
According to formulas (1) and (2), the
symbolic sequence of inbound tourist flows
to China is written in the form
{rDRrddrrdrdrdrrdddrrdrdrdrrrdd ... }. The
fluctuation patterns of inbound tourist flows
P
i
can be calculated by applying the following
formula:
Pi=(S3∗i−2,S3∗i−1,S3∗i).(3)
According to formula (3), the fluctuation pat-
terns of inbound tourist flows to China can be
expressed as {rDR,rdd,rrd,rdr,drr,ddd,rrd,
rdr,drr,rdd,... }. That is, the fluctuation pat-
terns evolve into each other with time {rDR
rdd rrd rdr drr ddd rrd
rdr drr rdd ... }. To identify the rule
of the transformation and detect significant
fluctuation patterns, we draw on complex
network theory to construct a weighted
network of inbound tourist flows to China.
The main idea of complex network theory is
946 Yongrui Guo et al.
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
to consider the relationships between various
parts of real complex systems as a complex
network (An et al., 2014). By analyzing the
structure of the network, we can better under-
stand the essential characteristics of the real
systems. In the symbolic sequence of fluctu-
ation patterns in inbound tourist flows to
China, the fluctuation patterns are defined as
nodes of the network, the transformations are
defined as edges, and the weight of an edge is
the number of transformations between the
two types of fluctuation patterns. The corre-
sponding network is shown in Figure 2.
Empirical Results
Important Fluctuation Patterns
We identified the significant fluctuation
patterns in the network of inbound tourist
flows using the degree and degree distribution
parameters of the complex network. The
concept of degree is the most fundamental
characteristic and measure of a node in a
network. The degree of a node in a complex
network is defined by the number of edges
directly connecting it to its neighbor. In this
paper, a node situated next to a given node is
considered its neighbor. The average degree
of a network is the average value of all node
degrees over the entire network. In undirected
networks, degree is a single number, but in
directed networks, nodes have two different
degrees, an in-degree and out-degree, corre-
sponding to the number of edges pointing
inward to and outward from those nodes. In
most cases, a node of higher degree is more
important than one of lower degree in a
network because it will have a more significant
influence on other nodes in the network in
terms of dynamics, information flows, and
data traffic, among other variables (Chen,
Wang, & Li, 2012). In this paper, the
Figure 2 The Weighted Network of Fluctuation Patterns in Inbound Tourist Flows to China.
Note: The greater the transition frequency, the thicker the line between nodes.
Modeling the Fluctuations of Tourist Flows 947
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
network of inbound tourist flows is a directed
network. The in-degree of the network of
inbound tourist flows represents other fluctu-
ation patterns transformed into a special
fluctuation pattern. The out-degree of the
network of inbound tourist flows represents
a special fluctuation pattern transformed into
other fluctuation patterns. Because the edge
is the sequential transformation between the
two nodes, the out-degree and in-degree of
each node are equal except for the first and
last nodes. In this paper, we use the out-
degree of the nodes to describe the degree
and degree distribution of the network of
inbound tourist flows. The degree distribution
of the network of inbound tourist flows can be
defined as follows:
p(k)=Ni
N,(4)
where N
i
denotes the number of nodes whose
degree equals kand Ndenotes the total
number of nodes in the network. A larger
out-degree of a special fluctuation pattern
implies a greater probability that this fluctu-
ation pattern will transform into another
pattern directly rather than through a series
of intermediate fluctuation patterns. A fluctu-
ation pattern of higher degree is more impor-
tant than one of lower degree.
We calculated the degree of every node in
the network of inbound tourist flows to
China. After that, we ranked the nodes in the
network from the highest to the lowest based
on the results degree calculation (Table 1).
According to Table 1, the first eight nodes
{rdr,RDr,RDd,RrD,rrd,DDR,drr,RDR}
have highest rank. The summed degree of
these nodes is 55.56%, and the degree of any
of these 8 nodes is more than 3% greater
than that of the others in the network of
inbound tourist flows, that is, 28.57% of the
nodes represent 55.56% of the degrees of the
network. This property means that these fluc-
tuation patterns have a significant role in
shaping macroscopic patterns of variations in
inbound tourist flows and in influencing
other patterns. In the process of transform-
ation of the fluctuation patterns, more trans-
formations should use these patterns as an
intermediate step. Identifying the out-degree
of nodes in the network of inbound tourist
flows contributes to understanding the fluctu-
ation rule and to forecasting future change.
For example, the fluctuation pattern RDd can
transform into three other types of patterns:
RrD,Rrd,andRdd. The transformation prob-
abilities are 0.57, 0.29, and 0.14, respectively.
The transformation probability between RDd
and RrD is greater than the other transform-
ation probabilities. The double-logarithmic
degree distribution and cumulative degree dis-
tribution plot of the network of inbound
tourist flows to China (Figure 3) shows few
nodes with a high degree value. Most of the
nodes are of lower degree value. Overall, the
degree distribution of the network of inbound
Table 1 Degree of Nodes in the Network of Inbound Tourist Flows to China
Node rdr RDr RDd RrD rrd DDR drr RDR ··· dDr
Degree 9 7 7 7 6 6 4 4 ··· 1
Rank 1 2 2 2 5 5 7 8 ··· 28
948 Yongrui Guo et al.
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
tourist flows to China follows a power-law dis-
tribution. Thus, only a handful of fluctuation
patterns significantly influence the fluctuations
in the inbound tourist flows to China.
Transformation Intermediary
In the transition process governing the fluctu-
ation patterns, we focus on the types of fluctu-
ation patterns that play an intermediary role in
the transformations. Thus, we can analyze the
betweenness centrality (BC) for each fluctu-
ation pattern in the network of inbound
tourist flows. In network theory, the between-
ness centrality of node vis the number of paths
from all nodes (except v) to all other nodes
that must pass through node v. The between-
ness centrality measures the intermediary or
middleperson power of a node. The between-
ness centrality of a node indicates its capability
to obtain and control resources or infor-
mation. Nodes with a high betweenness cen-
trality may have considerable influence
within a network by virtue of their control
over information or resources passing
between other nodes. The betweenness cen-
trality of a node vis given by the following
expression:
gv=
{i,j}
cij(v)
cij
,(5)
where g
v
is the betweenness centrality of a
node v,c
ij
is the total number of shortest
paths from node ito node j, and c
ij
(v) is the
number of those paths that pass through v.
The length of a path is the sum of the
weights of edges between iand j. The between-
ness centrality reveals the topological impor-
tance of nodes in its role in the transmission
of network information between each pair of
nodes. Therefore, the betweenness centrality
of a specific node can be explained as its
network influence. In the network of
inbound tourist flows, the network influence
of a specific node is the power of a fluctuation
pattern of inbound tourist flows to control or
affect other patterns in the network.
The betweenness centrality of the nodes in
the network of inbound tourist flows is
shown in Table 2. The differences among
nodes in betweenness centrality are evident.
The summed betweenness centrality of the
first 8 nodes is 63.18%, and the betweenness
centrality of any of these 8 nodes is more
Figure 3 Degree Distribution (a) and Cumulative Degree Distribution (b) in the Network of
Inbound Tourist Flows to China.
Modeling the Fluctuations of Tourist Flows 949
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
than 3% greater than that of the others; that
is, 28.57% nodes represent 63.18% of the
betweenness centrality of the network.
According to the statistics, six of the fluctu-
ation patterns {rdr,RDr,RDd,RrD,rrd,
DDR}listed in the top eight in degree rank
remain on the list in betweenness centrality.
This result means that these fluctuation pat-
terns are important intermediaries in the trans-
formation process for the fluctuation patterns
of inbound tourist flows. To a certain extent,
these fluctuation patterns can serve as a
precursor to the transformations between pat-
terns. These nodes are helpful in understand-
ing the inherent laws and transformation
information of the fluctuations in inbound
tourist flows to China. Furthermore, signifi-
cant differences among the betweenness cen-
trality of the nodes also suggest the presence
of higher volatility in the inbound tourist
flows to China.
Transformation Distance
Studying the shortest path length of the
network of inbound tourist flows can help us
understand the transition distance between
fluctuation patterns. The shortest path is the
minimum number of edges needed to connect
any two nodes. The average shortest path of
network Lis the average value of the shortest
path lengths of all of the connections between
two nodes, and it is defined as
L=2
N(N−1)
i≥j
dij,(6)
where d
ij
is the distance between nodes iand j
and Nis the total number of nodes in the
network. As shown in Table 3, the shortest dis-
tance and the longest distance between nodes
are 1 and 8, respectively. The most frequentdis-
tances, 3 and 4, represent more than 54% of the
cases in the network of inbound tourist flows.
The calculated value of the average shortest
path is 3.38. Therefore, if one type of fluctuation
pattern transforms into another, it will basically
change via three or four types of patterns.
Different types of fluctuation patterns rarely
Table 2 Betweenness Centrality of Nodes in the Network of Inbound Tourist Flows to China
Vertex rdr RDr RrD DDR DdR RDd rrd rDr ··· DdD
BC/% 14.71 11.53 7.46 7.45 6.81 6.75 4.65 3.78 ··· 0.32
Rank 1 2 3 4 5 6 7 8 ··· 28
Table 3 Frequencies of Shortest Path
Distances in the Network of Inbound Tourist
Flows to China
Distance Frequency Proportion (%)
1 60 7.9
2 138 18.3
3 213 28.2
4 199 26.3
5 100 13.2
6 36 4.8
7 9 1.2
8 1 0.1
950 Yongrui Guo et al.
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
transform into each other directly (distance of 1,
which accounts for 7.9% of cases). In the longest
path of transformation, a type of fluctuation
pattern transforms via 8 types of patterns, but
this case is infrequent (0.1% of cases). A trans-
formation occurred approximately every three
to four quarters. This information is used as a
basis for predicting changes in inbound tourist
flows in the next year. The results of the analysis
demonstrate that the transformations between
fluctuation patterns of inbound tourist flows
occur frequently.
Conclusions
This paper analyzed the fluctuation patterns of
monthly inbound tourist flows to China from
January 1990 to December 2012 using a
complex network approach. We constructed
a weighted network of inbound tourist flows
to China. The nodes represent 28 fluctuation
patterns of inbound tourist flows to China,
the edges are the transformations between
nodes, and the weight of an edge is the
number of transformations between the two
types of fluctuation patterns.
The most important nodes in the network are
rdr,RDr,RDd,RrD,rrd, and DDR. The
degree and betweenness of these nodes are
greater than those of other nodes in the
network of inbound tourist flows. The
summed degree and betweenness of these
nodes reach values of 46.67% and 52.58%,
respectively. In the process of transformation
of fluctuation patterns, more transformations
should use these patterns as an intermediate
step. We can identify significant fluctuation
patterns of inbound tourist flows using these
topologically important nodes in the network.
These significant fluctuation patterns of
inbound tourist flows play a key role in
pattern transformation and can be viewed as
the prelude to changes in inbound tourist
flows. These fluctuation patterns are helpful
in understanding the inherent laws and trans-
formation information related to the fluctu-
ation in inbound tourist flows to China. The
average transition distance was 3.38, and a
transformation occurred approximately every
3–4 quarters. These results demonstrate that
the transformations between fluctuation pat-
terns of inbound tourist flows occur frequently.
This paper analyzed the complex character-
istics of the fluctuation patterns of monthly
inbound tourist flows to China from the perspec-
tive of network topology. This method was of
guiding significance in identifying the important
fluctuation patterns and understanding the
inherent laws of the fluctuations in tourist
flows. The statistical properties of fluctuations
in tourist flows are important for modeling the
complex dynamics of tourist flows and are of
great significance for practical applications such
as tourist flow risk estimation and tourism flow
forecasting. According to this method and the
results in this paper, a forecasting model can be
built using the transformation intermediary, the
transformation probability and the transform-
ation time. This model differs from previous
tourism demand forecasting models because the
model is based on the fluctuation patterns of
tourist flows but not on the time series itself.
This difference is a direction for future research.
Acknowledgement
This work was supported by the National
Natural Science Foundation of China (Grant
No. 41171121, 41301134).
References
An, H., Gao, X., Fang, W., Huang, X., & Ding, Y. (2014).
The role of fluctuating modes of autocorrelation in
Modeling the Fluctuations of Tourist Flows 951
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
crude oil prices. Physica A: Statistical Mechanics and its
Applications,393, 382– 390.
Assaf, A. G., Barros, C. P., & Gil-Alana, L. A. (2011).
Persistence in the short- and long-term tourist arrivals to
Australia. Journal of Travel Research,50(2), 213– 229.
Athanasopoulos, G., & de Silva, A. (2012). Multivariate
exponential smoothing for forecasting tourist arrivals.
Journal of Travel Research,51(5), 640– 652.
Bonanno, G., Caldarelli, G., Lillo, F., Micciche
´, S., Vande-
walle, N., & Mantegna, R. N. (2004). Networks of equi-
ties in financial markets. The European Physical Journal
B – Condensed Matter and Complex Systems,38(2),
363– 371.
Chan, F., Lim, C., & McAleer, M. (2005). Modelling
multivariate international tourism demand and vola-
tility. Tourism Management,26(3), 459– 471.
Chang, Y. W., & Liao, M. Y. (2010). A seasonal ARIMA
model of tourism forecasting: The case of Taiwan. Asia
Pacific Journal of Tourism Research,15(2), 215– 221.
Chen, G. R., Wang, X. F., & Li, X. (2012). Introduction to
complex networks: Models, structures and dynamics.
Beijing: Higher Education Press.
Cho, V. (2003). A comparison of three different
approaches to tourist arrival forecasting. Tourism Man-
agement,24(3), 323– 330.
Cho, V. (2009). A study on the temporal dynamics of
tourism demand in the Asia Pacific region. International
Journal of Tourism Research,11(5), 465– 485.
Chu, H. P., Yeh, M. L., & Chang, T. Y. (2014). Are
visitor arrivals to China stationary? An empirical note.
Asia Pacific Journal of Tourism Research,19(2), 248–
256.
CNTA. (2013). The yearbook of China tourism statistics
2013. Beijing: China Tourism Press.
Coshall, J. T. (2006). Time series analyses of UK outbound
travel by air.Journal of Travel Research,44(3), 335– 347.
Cuccia, T., & Rizzo, I. (2011). Tourism seasonality in cul-
tural destinations: Empirical evidence from Sicily.
Tourism Management,32(3), 589– 595.
De Cantis, S., Ferrante, M., & Vaccina, F. (2011). Seaso-
nal pattern and amplitude – a logical framework to
analyse seasonality in tourism: An application to bed
occupancy in Sicilian hotels. Tourism Economics,
17(3), 655– 675.
Fourie, J., & Santana-Gallego, M. (2011). The impact of
mega-sport events on tourist arrivals. Tourism Manage-
ment,32(6), 1364– 1370.
Gil-Alana, L. A. (2005). Modelling international monthly
arrivals using seasonal univariate long-memory pro-
cesses. Tourism Management,26(6), 867– 878.
Goh, C., & Law, R. (2002). Modeling and forecasting
tourism demand for arrivals with stochastic nonstation-
ary seasonality and intervention. Tourism Management,
23(5), 499– 510.
Gursoy, D., Parroco, A. M., & Scuderi, R. (2013). An
examination of tourist arrivals dynamics using short-
term time series data: A space-time cluster approach.
Tourism Economics,19(4), 761– 777.
Gustavsson, P., & Nordstrom, J. (2001). The impact of
seasonal unit roots and vector ARMA modelling on
forecasting monthly tourism flows. Tourism Econ-
omics,7(2), 117– 133.
Hadavandi, E., Ghanbari, A., Shahanaghi, K., & Abba-
sian-Naghneh, S. (2011). Tourist arrival forecasting by
evolutionary fuzzy systems. Tourism Management,
32(5), 1196– 1203.
Higham, J., & Hinch, T. (2002). Tourism, sport and
seasons: The challenges and potential of overcoming
seasonality in the sport and tourism sectors. Tourism
Management,23(2), 175– 185.
Jang, S. C. (2004). Mitigating tourism seasonality: A
quantitative approach. Annals of Tourism Research,
31(4), 819– 836.
Kim, J. H., & Moosa, I. A. (2001). Seasonal behaviour of
monthly international tourist flows: Specification and
implications for forecasting models. Tourism Econ-
omics,7(4), 381– 396.
Kuo, H., Liu, K. E., & Chen, C. (2014). Modeling Japa-
nese tourism demand for Asian destinations: A
dynamic AIDS approach. Asia Pacific Journal of
Tourism Research,19(1), 86– 102.
Lacasa, L., Luque, B., Ballesteros, F., Luque, J., & Nuno,
J. C. (2008). From time series to complex networks: The
visibility graph. Proceedings of the National Academy
of Sciences of the United States of America,105(13),
4972– 4975.
Lean,H.H.,&Smyth,R.(2009).Asianfinancialcrisis,
avian flu and terrorist threats: Are shocks to Malay-
sian tourist arrivals permanent or transitory? Asia
Pacific Journal of Tourism Research,14(3),
301–321.
Li, G., Song, H., & Witt, S. F. (2005). Recent develop-
ments in econometric modeling and forecasting.
Journal of Travel Research,44(1), 82– 99.
Li, G., Wong, K. K., Song, H., & Witt, S. F. (2006).
Tourism demand forecasting: A time varying parameter
error correction model. Journal of Travel Research,
45(2), 175– 185.
Li, P., & Wang, B. H. (2006a). An approach to Hang Seng
Index in Hong Kong stock market based on network
952 Yongrui Guo et al.
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015
topological statistics. Chinese Science Bulletin,51(5),
624–629.
Li, P., & Wang, B. H. (2006b). A dynamic model of Hang
Seng Index based on complex network eigenvectors.
Systems Engineering,24(3), 73– 77.
Li, P., & Wang, B. H. (2007). Extracting hidden fluctu-
ation patterns of Hang Seng stock index from network
topologies. Physica A: Statistical Mechanics and its
Applications,378(2), 519– 526.
Lim, C., & McAleer, M. (2001). Monthly seasonal vari-
ations – Asian tourism to Australia. Annals of
Tourism Research,28(1), 68– 82.
Nadal, J. R., Font, A. R., & Rossello, A. S. (2004). The
economic determinants of seasonal patterns. Annals of
Tourism Research,31(3), 697– 711.
Shareef, R., & McAleer, M. (2005). Modelling inter-
national tourism demand and volatility in small island
tourism economies. International Journal of Tourism
Research,7(6), 313– 333.
Song, H., & Li, G. (2008). Tourism demand modelling
and forecasting – a review of recent research. Tourism
Management,29(2), 203– 220.
Song, H., Li, G., Witt, S. F., & Athanasopoulos, G.
(2011). Forecasting tourist arrivals using time-varying
parameter structural time series models. International
Journal of Forecasting,27(3), 855– 869.
Song, H., Lin, S. S., Zhang, X. Y., & Gao, Z. X. (2010).
Global financial/economic crisis and tourist arrival
forecasts for Hong Kong. Asia Pacific Journal of
Tourism Research,15(2), 223– 242.
Song, H., & Witt, S. F. (2006). Forecasting international
tourist flows to Macau. Tourism Management,27(2),
214–224.
Tao, W., & Ni, M. (2010). Study on the comparison of
tourism demand forecast between China and western
countries: Basic theory and models. Tourism Tribune,
25(8), 12– 17.
Turner, L. W., & Witt, S. F. (2001). Forecasting
tourism using univariate and multivariate structural
time series models. Tourism Economics,7(2), 135–
147.
Vergori, A. S. (2012). Forecasting tourism demand:
The role of seasonality. Tourism Economics,18(5),
915–930.
Wong, K. K., Song, H., Witt, S. F., & Wu, D. C. (2007).
Tourism forecasting: To combine or not to combine?
Tourism Management,28(4), 1068– 1078.
Yan, M., & Wall, G. (2003). Disaggregating visitor flows:
The example of China. Tourism Analysis,7(3–4), 191 –
205.
Yang, Y., & Wong, K. K. F. (2013). Spatial distribution of
tourist flows to China’s cities. Tourism Geographies,
15(2), 338– 363.
Yang, Y., Pan, B., & Song, H. (2014). Predicting hotel
demand using destination marketing organization’s
web traffic data. Journal of Travel Research,53(4),
433–447.
Zhang, J. H., Zhou, H. X., Jiang, L., & Wang, Y. G.
(2010). Network topologies of Shanghai stock index.
Physics Procedia,3(5), 1733– 1740.
Modeling the Fluctuations of Tourist Flows 953
Downloaded by [Stephen F Austin State University] at 04:23 05 August 2015