Experience mining based innovative method for promoting urban sustainability

Accepted Manuscript Experience mining based innovative method for promoting urban sustainability

Liyin Shen, Hang Yan, Xiaoling Zhang, Chenyang Shuai PII:

S0959-6526(17)30792-8

DOI:

10.1016/j.jclepro.2017.04.074

Reference:

JCLP 9429

To appear in:

Journal of Cleaner Production

Received Date:

15 October 2016

Revised Date:

08 April 2017

Accepted Date:

09 April 2017

Please cite this article as: Liyin Shen, Hang Yan, Xiaoling Zhang, Chenyang Shuai, Experience mining based innovative method for promoting urban sustainability, Journal of Cleaner Production (2017), doi: 10.1016/j.jclepro.2017.04.074

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Highlights 

The parameter of similarity is introduced to measure the differences between cities



Six urban features are selected to describe the major characteristics of a city.



A hybrid method is developed to measure the similarity between two cities.



A case study is used to verify the application of the hybrid method.

ACCEPTED MANUSCRIPT 1

Experience mining based innovative method for promoting urban

2

sustainability Liyin Shen1; Hang Yan2*; Xiaoling Zhang3 and Chenyang Shuai4

3 4

1

Professor, School of Construction Management and Real Estate, International

5

Research Centre for Sustainable Built Environment, Chongqing University, China,

6

[email protected]

7

2*

Corresponding Author, PhD Researcher, School of Construction Management

8

and Real Estate, International Research Centre for Sustainable Built Environment,

9

Chongqing University, Chongqing, China. [email protected]

10 11

3

Associate Professor, Urban Research Group, City University of Hong Kong;

Hong Kong, [email protected]

12

4 Master student, School of Construction Management and Real Estate, International

13

Research Centre for Sustainable Built Environment, Chongqing University, Chongqing,

14

China. [email protected]

15 16 17 18 19 20 21

ACCEPTED MANUSCRIPT 1

Abstract

2

The promotion of sustainable urbanization has generated a growing number of best

3

practice cases, which has led to the development of a prototype of an Experience

4

Mining System (ExMS) for better use of the valuable experience embodied in these

5

best practices. This paper presents an innovative method for the effective application

6

of the experience mining system thus to promote the urban sustainability. It is

7

considered essential to appreciate the urban features when ExMS is used to mine

8

effective experiences for improving the sustainability of a specific urban area. For this

9

purpose, this paper introduces the measure of similarity between a concerned case and

10

those cases reported in an ExMS database in order to ensure the effectiveness of mining

11

good experience. The method presented in this paper, it helps to find out useful

12

experiences from existing best sustainable urbanization practices by considering the

13

similarity between the case concerned and the sample best practices. The similarity is

14

measured from six perspectives of urban features, including landform, climate, urban

15

scale, development level, Gini coefficient, and GDP performance. The value formats

16

of these six urban features include crisp symbol, crisp number, interval numbers, and

17

fuzzy linguistic variable. A hybrid similarity measure will therefore guide the process

18

in mining effective experiences for the promotion of urban sustainability.

19

Key words: Innovative method; Urban sustainability; Experience mining system

20

(ExMS); Best practices; Similarity measure; Urban feature

21

ACCEPTED MANUSCRIPT 1

1. Introduction

2

Urbanization, which is defined as a movement of people from rural to urban areas,

3

has been one of the most prominent trends of the 20th and 21st century (United Nations,

4

2010). Globally, more than half of world’s population (54 percent) now resides in urban

5

areas, up from 30% in 1950. It is predicted that this proportion will reach to 66% by

6

2050 (United Nations, 2014). Developing countries such as China and India are

7

experiencing a faster urbanization process than other regions. (Zhang, 2015; Zhai,

8

2014; Zhan et al, 2016; Gan et al, 2016). Urbanization as the main driving force of

9

social development brings a lot of benefits, such as diversity, jobs, education and

10

improvement of residents’ living quality (Dye, 2008; Dyson, 2011). However, rapid

11

urbanization in developing countries has also caused a variety of problems such as air

12

and water pollution, global warming, progressive exhaustion of resources, inefficient

13

consumption of energy, insufficient housing, and health problem (Yang et al, 2016; Liu

14

et al, 2014; Gan et al., 2015; Wu et al., 2016; Shuai et al., 2017). Promoting the

15

sustainability performance in urbanization process is recognized as an important

16

approach to mitigate these problems, which pursues the balance between socio-

17

economic and environment conservation (Liu et al., 2016; Wang et al., 2016, Ernst et

18

al., 2016; Krajnc and Glavič, 2005). The European Commission (2006) pointed out that

19

the challenge in gaining urban sustainability is how to solve both the problems

20

experienced within cities and the problems caused by cities.

21

In line with the promotion of sustainable urbanization, international institutions

ACCEPTED MANUSCRIPT 1

and local governments throughout the world have been devoting great amount of efforts

2

in practicing sustainable urbanization at different levels. For example, United Nations

3

Human Settlements Programme (UN-Habitat), United Nations Environment

4

Programme (UNEP), and the World Bank have been promoting various sustainable

5

urbanization schemes, such as the Sustainable Cities Programme, Millennium

6

Declaration, the 10-year framework of programmes on sustainable consumption and

7

production patterns. In respond, governments at national and city levels have also

8

developed some programmes related to sustainable urbanization. For example, “the

9

Low-carbon City” initiative is introduced by National Development and Reform

10

Committee (NDRC) in China for tackling environmental degradation and pursuing

11

sustainable urban development (Li and Qiu, 2015). In order to achieve a sustainable

12

urban development, the government of Mexico City launched a 15-year program, called

13

Mexico City Green Plan, which is composed of seven subprograms: land conservation,

14

public spaces, air pollution, waste management and recycling, water supply and

15

sanitation, transportation, and mobility (Cheller et al., 2015). Whilst urban

16

sustainability is contributed by various sectors of a city, building sector plays a specially

17

important role. The sustainable performance of building stock has significant impact on

18

urban sustainability (Shi et al, 2013). It is widely appreciated that building sector is one

19

of the largest energy consumers and carbon emitters. According to the report of

20

Intergovernmental Panel on Climate Change, building sector contributes to

21

approximately 32% of final energy consumption and 19% of energy-related GHG

ACCEPTED MANUSCRIPT 1

emissions (IPCC, 2015). In line with this, many green building assessment tools have

2

been developed to promote the practices of green building, such as Leadership in

3

Energy and Environment Design (LEED), Building Research Establishment

4

Environment Assessment Method (BREEAM), and Green Building Label (Zuo and

5

Zhao, 2014). These efforts have produced a large amount of best experiences in

6

practicing sustainable urbanization. All these sustainable initiatives, schemes and

7

polices have generated numerous best practices. The value of these practices is

8

immeasurable, and the lessons learned from these practices could provide direct

9

references on how to manage and resolve the current problems which may be similar

10

to what experienced in the past. In order to reuse these successful experience in an

11

effective way, an Experience-mining System (ExMS) for sharing successful experience

12

in promoting sustainable urbanization has been introduced by Shen et al (2013).

13

ExMS is a new methodology that extracts valuable experiences from past practical

14

cases as decision-making references to solve new problems in promoting sustainability

15

of urbanization. The methodology has three major components, namely, a Sustainable

16

Urbanization Practices Database (SUPD), a Refinery process, and a Mine-sweeper.

17

Among these components, Mine-sweeper assumes a more complicated function in

18

searching for the most similar cases from SUPD to solve the new problems (Shen et al.,

19

2013). Case retrieval process is to ensure that the mined cases in the database are very

20

similar to a target problem. Previous studies appreciate that similarity between cases

21

plays the key role to ensure the effectiveness of using past experiences as references in

ACCEPTED MANUSCRIPT 1

solving new problems (Behbahani et al., 2012; Pereira & Madureira, 2013; Perner,

2

2014). However, case retrieval process in the existing ExMS only considers problem

3

nature. It does not consider the level of similarity of the urban backgrounds between

4

the target problem and those to be mined from SUPD. These background differences

5

will lead to different results even though same solutions are adopted to similar

6

problems. Generally, city governments intend to refer the successful experiences gained

7

in developed regions when they look for solutions to solve problems in the urbanization

8

process. However, cities around world are different in many aspects, such as

9

demographic, geographic, economic, sociological, cultural, and others (Tadic et al.,

10

2014). Thus, experience gained from other cities should not be adopted indiscriminately

11

by decision makers. Therefore, this study proposes an innovative method by adding a

12

mechanism of incorporating the similarity of urban features on the existing method of

13

ExMS to increase the effectiveness of the methodology in promoting urban

14

sustainability. This mechanism can assist government stakeholders to find the most

15

appropriate experience from other cities which have similar urban backgrounds in

16

addressing unsustainable problems such as haze, congestion, social inequality, and

17

resource shortage.

18

2. Literature review

19

Many researchers suggested using experience from developed regions to solve similar

20

problems encountered during urbanization in developing areas. Li et al. (2011)

21

indicated that the experience of urban planning and ecological development in Canada

ACCEPTED MANUSCRIPT 1

provides a valuable reference for China to mitigate the climate change and reduce

2

carbon emissions. Lehmann (2013) compared the experiences from two cases about

3

urban development patterns for new sub-centers between Berlin (Germany) and

4

Shanghai (China). The experience from Berlin is recommended as a reference for

5

designing and developing a sub-center in Shanghai. Yin et al. (2014) discussed the

6

effects of the high-speed railway (HSR) network on urban development and the

7

planning challenges for China by translating international experiences into Chinese

8

context. In order to promote the development of low-carbon city, the Chinese

9

government launched a demonstration program of 5 pilot provinces and 8 pilot cities in

10

2010. The experience from these pilot cities have been investigated and recommended

11

for extension to the whole country (Khanna et al., 2014; Flynn et al., 2016). However,

12

it appears that little existing study has addressed how to incorporate urban background

13

when exploring experiences from other cities. Without this incorporation, the

14

experiences explored may not be adaptable to solve effectively a target problem which

15

is under a very different urban background.

16

Others researchers have highlighted the importance of considering urban

17

background and features when identifying solution for sustainable urbanization. Yu

18

(2014) pointed out that implementation of sustainable city development should be

19

closely associated with urban features and challenges. In appreciating the close

20

relationship between urban sustainability and green building, Boschman and Gabriel

21

(2013) emphasized that the regional characteristics should be recognized when

ACCEPTED MANUSCRIPT 1

promoting the best practices of green building. Urban features can be examined from

2

various perspectives. In discussing how the experiences of cycling activities can be

3

promoted in urban areas, Meng et al. (2014) referred to four urban features, namely,

4

city function, land area, population, purchasing power, and economic. Russo and Comi

5

(2012) discussed the experiences in formulating the goals of environment sustainability

6

by analyzing the effects of density of population, which is a main urban characteristic.

7

Zhao et al. (2014) considered four urban features, namely, population, job, income per

8

capital, and the government expenditure, when identifying the potential factors that

9

affect the use of public bike. Schmidt et al. (2015) compared the urbanization patterns

10

in Moshi and Dar Es Salaam, Tanzania from four urban features including

11

demographic, economic, environment and institutional perspectives. To compare the

12

Gold Coast with other major cities in Australia, Dedekorkut-Howes and Bosman (2015)

13

applied a number of urban features including historical evolution, urban development

14

and form, development and demographic structures, population growth, and economic

15

indicators.

16

The literature review appreciates that different researchers adopted different urban

17

features to describe city background because they focus on different urban problems. It

18

is considered important to incorporate the urban features when identifying existing

19

experiences for decision making references. In other words, the similarity of urban

20

features between the existing experience and a target problem should be considered.

21

However, there is little existing study in addressing how to analyze the similarity of

ACCEPTED MANUSCRIPT 1

urban features. Therefore, this study is designated to find an approach of addressing the

2

similarity of urban features in using the methodology of ExMS for promoting

3

sustainable urbanization.

4

3. Research methods

5

Two major research methods are used in combination. Literature review is

6

conducted for selecting effective urban features and their measurement indicators. A

7

hybrid method is used to analyze the similarity of urban features between different

8

urban backgrounds.

9

A number of previous studies have discussed urban features from various

10

perspectives. These literature works provide the foundation for choosing effective

11

urban features. And three types of literature works are accessed: (1) research

12

publications on the topics relating to urban features; (2) common search engine, such

13

as, Wikipedia, for searching typical cities in order to gain general understanding on

14

urban features; (3) indicator systems for sustainable urbanization, such as the indicator

15

system established by The World Bank, and Millennium Development Goals Indicators

16

established by United Nations.

17

Establishing an effective similarity measure is crucial to retrieve useful

18

experiences in solving a new problem. In previous studies, similarity measure has been

19

used to retrieve the similar historical cases in various disciplines such as medical

20

diagnosis (Guessoum et al., 2014, EI-Fakdi et al., 2014), manufacturing industry (Kuo,

21

2010), business (Carmona et al., 2013), etc.

ACCEPTED MANUSCRIPT 1

Typical methods for calculating similarity include distance-based method, soft

2

computing method, and hybrid method. The distance-based method measures similarity

3

or distance between features with crisp numbers (Hjaltason and Samet, 2003). There

4

are two types of distance-based methods to assess similarity level between a pair of

5

features or cases, namely, the Euclidean distance and the Hamming distance (Liao,

6

2004; Chang et al, 2005). However, in the urban context, urban features are often

7

described by using fuzzy terms such as large city, fast economic development, and high

8

level welfare society. In this case, the distance-based method cannot be effectively used

9

to measure these fuzzy features. Therefore, soft computing method is adopted to

10

address these urban features which are described in linguistic terms. (Xiong, 2011;

11

Slonim and Schneider, 2001). In fact, in addition to crisp number and fuzzy term, other

12

formats of value include interval value and crisp symbol (Fan, et al, 2015). For example,

13

the GDP per capital in a city can be described by an interval value, such as between

14

$5000 to $6000. Climate can be expressed in a crisp symbol, such as humid continental

15

climate.

16

Having realized that there are four value formats in describing urban features, a

17

hybrid similarity measure method is adopted in this study to measure the similarity

18

between urban features. There are existing hybrid similarity measures are developed

19

from different perspectives, for example:

20



21

the similarity between crisp number feature and interval number feature (Liu & Xi, 2011),

ACCEPTED MANUSCRIPT 1



2 3

(Zarandi et al., 2011), 

4 5

the similarity between crisp symbol feature and crisp number feature (Castro et al., 2009; Kong et al., 2013);



6 7

the similarity between interval value feature and interval and fuzzy number feature

the similarity measure for comparing cases which have a mixture of crisp and fuzzy features (Liao et al, 1998),



the similarity measure for describing five formats of attribute values in the

8

application of case-based reasoning principle (Fan et al, 2014).

9

The above discussions reveal that hybrid similarity measures are needed in

10

measuring the similarity between urban features which are in various value formats.

11

This study considers four formats of urban feature value, namely, crisp symbol, crisp

12

number, interval number, and fuzzy linguistic variable.

13

4. Urban features and their formats

14

“The city designates the space produced by the interaction of historically and

15

geographically specific institutions, social relations of production and reproduction,

16

practices of government, forms and media of communication, and so forth” (Donald,

17

1992). As the focus of demographic and economic, city hosts the majority of the human

18

population (Flynn et al., 2016). It is an intricate system that involves a large set of

19

interrelated components across environment, economic, social, and other dimensions

20

(Landry, 2006, Mutisys and Yarime, 2014). Therefore, the multidimensionality of city

21

leads to difficulty in describing the background of city. Many scholars has adopted

ACCEPTED MANUSCRIPT 1

various urban features or indicators to describe a city from different aspects (Schmidt

2

et al., 2015; Shen et al., 2015). Based on the literatures review in section 2 of this study,

3

six typical urban features are selected from three perspectives economic,

4

environmental, and social, where landform and climate represent the environment

5

perspective, urban scale and development level represent the social aspect, Gini

6

coefficient and GDP performance represent the economics perspective. Each type of

7

urban feature is measured by certain indicators. For instance, urban scale is described

8

by the total population. Landform is indicated by hills, mountains, plains, and plateaus.

9

These commonly adopted urban features and their indicators are summarized in Table

10

1.

11 12

(Insert Table 1 here)

13 14

Landform (C1)

15

As shown in Table 1, the feature “landform” for a specific city is indicated as hills,

16

mountains, plateaus, or plains. Each type of landform is characterized by its slope,

17

elevation, soil and rock type, stratification and orientation, as summarized in Table 2.

18 19 20 21

(Insert Table 2 here)

ACCEPTED MANUSCRIPT 1

Climate (C2)

2

Climate is an important feature for describing the background for a specific city.

3

The commonly used climate classification system was introduced by Köppen (1900),

4

in which monthly temperature and precipitation are used to define the boundaries of

5

different climate types. The Köppen climate classification consists of five major groups

6

and various sub-groups, as shown in Table 3:

7 8

(Insert Table 3 here)

9 10

Urban scale (C3)

11

“Total population” is widely adopted to describe urban scale, which is typically

12

classified into five categories (Economic Development council of Small and Medium-

13

sized Cities in China Society of Urban Economy, 2010):

14



Small city: population is less than 500,000

15



Medium-size city: population is between 500,000 and 1,000,000

16



Larger city: population is between 1,000,000 and 3,000,000

17



Super city: population is between 3,000,000 and 10,000,000

18



Mega-city: population is above 10,000,000

19 20 21

Development level (C4) Development level for a particular city can be indicated by urbanization level to

ACCEPTED MANUSCRIPT 1

reflect the shift of population from rural to urban areas. (United Nations, 2010).

2

According to Northam Curve (Northam, 1979), urbanization is classified into three

3

stages: early stage when the level of urbanization rate is below 30%, middle stage when

4

the urbanization rate is between 30% and 70%, and mature stage when the urbanization

5

rate is above 70%.

6 7

Gini coefficient (C5)

8

Gini coefficient measures the extent to which the distribution of income or

9

consumption expenditure among individuals or households within an economy deviates

10

from a perfectly equal distribution (The World Bank, 2015). Some researchers

11

suggested adopting Gini coefficient to evaluate the social development level (Atkinson,

12

2002; Shen et al., 2017). Based on Gini coefficient, social income distribution can be

13

classified into five groups (Gao, 1995):

14



Strong equality: Gini coefficient is less than 0.2.

15



Relative equality: Gini coefficient is between 0.2 and 0.3.

16



Reasonable equality: Gini coefficient is between 0.3 and 0.4

17



Relative inequality: Gini coefficient is between 0.4 and 0.5

18



Strong inequality: Gini coefficient is above 0.5

19

GDP performance (C6)

20

In general, GDP performance of a city is measured by GDP per capita (United

21

Nations Department of Economic, 2007; Moussiopoulos et al., 2010; Shen et al., 2016).

ACCEPTED MANUSCRIPT 1

In accordance with the classification criteria by the International Monetary Fund, the

2

level of GDP per capital is classified into five categories:

3



First level: GDP per capita is under $500

4



Second level: GDP per capita is between $500 and $2500

5



Third level: GDP per capita is between $2500 and $10000

6



Fourth level: GDP per capita is between $10000 and $25000

7



Fifth level: GDP per capita is above $25000

8 9

Value formats of urban features

10

The six urban features discussed above assume four types of value formats: crisp

11

symbol, crisp number, interval number, and fuzzy linguistic variable.

12



“Crisp symbol” is used to express the terms with definite meanings. For

13

example, the feature “Landform” can be expressed by “hills”, “plains”,

14

“mountains”, and “plateaus”.

15



16 17

“Crisp number” is used to describe a certain value. For example, the value of “GDP per capital” is “$2000”.



“Interval number” is used to express the range of an indicator value, indicating

18

that the concerned urban feature changes in different time periods. For

19

example, the urbanization rate can be described as between 45% and 50% over

20

a period of time.

21



“Fuzzy linguistic variable” is a natural language expression or word that

ACCEPTED MANUSCRIPT 1

describes imprecisely an attribute (Pal & Shiu, 2004). For example,

2

urbanization development status can be described by such linguistic terms as

3

“early”, “medium” and “mature stage”.

4 5

5. Hybrid similarity measure

6

As appreciated early in the introduction section, this study aims to find the level

7

of similarity between a target city where there is a specific sustainability related

8

problem to be solved and a reported city which has produced successful experience

9

stored in a case database. The discussion in previous section on urban features provides

10

the basis to establish the measure of the similarity.

11

The similarity between a reported city and a target city will be analyzed under the

12

following seven scenarios:

13 14 15 16

A. Scenario One (S1): the urban features for both the reported city and target city are described in crisp symbol; B. Scenario Two (S2): the urban features for both the reported city and target city are described in crisp numbers;

17

C. Scenario Three (S3): the urban feature for the reported city (or target city) is

18

described in crisp numbers, and that for the other one is described in interval

19

numbers;

20

D. Scenario Four (S4): the urban feature for the reported city (or target city) is

21

described in crisp numbers, and that for the other one is described in fuzzy

ACCEPTED MANUSCRIPT 1 2 3

linguistic variables; E. Scenario Five (S5): the urban features for both the reported city and target city are described in interval numbers;

4

F. Scenario Seven (S6): the urban feature for stored city (or target city) is

5

described in interval numbers, and that for the other one is described in fuzzy

6

linguistic variables;

7

G. Scenario Six (S7): The urban features for both the reported city and target city

8

are described in fuzzy linguistic values;

9

These scenarios can be expressed graphically in Figure 1.

10 11

(Insert Figure 1 here)

12 13 14 15 16 17 18 19

For establishing similarity measures under these seven scenarios, the following background conditions are defined: The set Z  {Z1 , Z 2 ,...., Z m } is a set of

m

reported cities in an experience database,

Z i denotes the ith city, i  M (1, 2,..., m) , Z 0 denotes the target city; The set C  (C1 , C2 ,..., C6 ) is urban feature set, where C j denotes the value of the

jth feature,

j  N (1, 2,..., 6) ;

 W  ( w1 , w2 ,..., w6 )T is the vector of weighting values between six urban feature,

20

subject to:

21

feature C j .



jN

w j  1 , and 0  w j  1 ; where w j denotes the weighting value of the

ACCEPTED MANUSCRIPT 1 2 3 4

 Pi  ( pi1 , pi 2 ,..., pi 6 )T is the vector of urban feature values, where pij denotes the value of the feature C j in referring to the reported city Z i ;

 P0  ( p01 , p02 ,..., p06 )T is the vector of the urban feature values, where p0 j denotes the value of feature C j in referring to the target city Z 0 .

5 6

(1) Similarity measure in referring to scenario one (S1)

7

In scenario one, both a reported city and the target city apply a crisp symbol for

8

describing the feature C j . As a crisp symbol is used to express terms with definite

9

meanings, it is suggested to measure the similarity between Z i and Z 0 in regard to the

10

feature C j , denoted as S1i ( Z i , Z 0 ) , by judging whether or not the value of the crisp

11

symbol carried by Z i is equivalent to that carried by Z 0 (EI-Fakdi et al., 2014):

12 13

1 S1i ( Z i , Z 0 )   0

if P0 j  Pij if P0 j  Pij

iM, jN

(1)

(2) Similarity measure in referring to scenario two (S2)

14

In the circumstance of scenario two, both a reported city and the target city assume

15

a crisp number for describing the urban feature C j . It is considered effective to measure

16

the similarity between Z i and Z 0 in referring to the feature C j by calculating the

17

distance between the values of the two crisp numbers, denoted as S 2i ( Z 0 , Z i ) (Castro

18

et al., 2009). The shorter the distance, the more similar the reported city is to the target

19

city. This is the distance-based method for obtaining the similarity between a reported

20

city Z i and the target city Z 0 .

21

According to Liao et al (1998), S 2i ( Z 0 , Z i ) is obtained from the follows:

ACCEPTED MANUSCRIPT S 2i ( Z 0 , Z i )  1  D( pij , p0 j ), i  M , j  N

1 2 3

(2)

Where D( pij , p0 j ) is the Humming distance between pij and p0 j which can be obtained through the following formula:

4

D ( pij , p0 j ) 

pij  p0 j

pij , p0 j  [ ,  ]

 -

 m  Min p ij   i  p0 j 

5

( when p ( when p

 m p ij  Max i   p  0j

6

( when p ( when p

iM, jN m

0 j  Min p ij ) i

m

0j

 Min p ij ) i

m

0j

 Max p ij ) i

m

0 j  Max p ij ) i

7 8

(3)Similarity measure in referring to scenario three (S3)

9

In this scenario, there are two possibilities exist: (a) Z i assumes a crisp number

10

and Z 0 assumes an interval number for describing the urban feature C j ; (b) Z 0 assumes

11

an interval number and Z 0 assumes a crisp number for describing the urban feature C j .

12

In referring to the possibility (a), the crisp number of the urban feature C j for Z i

13

ⅠⅡ is pij and that the interval value of C j for Z 0 is p0 j  [ p0 j , p0 j ] . According to Slonim

14

& Schneider (2001), the similarity between Z i and Z 0 is obtained from the following

15

model:

16

S 3i ( Z 0 , Z i )  1-



pⅡ 0j pⅠ 0j

x  Pij dx

(   （） ) pⅡⅠ 0 j  p0 j

17

For the possibility (b), the model (3) is also applicable.

18

Where α and β have been defined in addressing the scenario two.

(3)

ACCEPTED MANUSCRIPT 1

(4) Similarity measure in referring to scenario four (S4)

2

This scenario has two possibilities: (a) Z i assumes a crisp number, and Z 0

3

assumes a fuzzy linguistic variable for describing the urban feature C j ; (b) Z i assumes

4

a fuzzy linguistic variable and Z 0 assumes a crisp number for describing the feature

5

Cj .

6

In referring to the possibility (a), a fuzzy membership function is introduced to express

7

the linguistic term. Determination of fuzzy membership functions is subjective in nature

8

and depends on the specific circumstance. According to previous study (Pal & Shiu,

9

2004), triangular membership function and trapezoid membership function are the most

10

common fuzzy membership functions. In this study, trapezoidal membership function

11

is used to describe “urbanization level”, such as “early stage”, “middle stage”, and

12

“mature stage”. These three levels of urbanization have been addressed early in this

13

paper when discussing the urban feature “Development level (C4)”, which can be

14

described graphically in Figure 2. The three corresponding fuzzy membership

15

functions, namely, early ( x) ,

middle ( x) , and mature ( x) can be constructed as follows:

16 17

(Insert Figure 2 here)

18 19

20

1  0.4  x  early ( x)    0.4  0.2 0

x  0.2 0.2  x  0.4 0.4  x

ACCEPTED MANUSCRIPT

1

2

x  0.2

0  x  0.2   0.4  0.2  middle ( x)  1  0.8  x   0.8  0.6 0

0.2  x  0.4 0.4  x  0.6 0.6  x  0.8 0.8  x

x  0.6

0  x  0.6  mature ( x)    0.8  0.6 1

0.6  x  0.8 0.8  x

3 4 5

In referring to these fuzzy membership functions, the similarity between Z i and

6

Z 0 , denoted by S 4（ i Z i , Z 0 ) , can be measured by using the following formula, as

7

proposed by Pal and Shiu (2004): 

  ( x) p  x dx S 4（Z , Z )  1-   (    )   ( x)dx  ij

8

i

i

0

(4)

9

where  ( x) is the membership functions of a specific linguistic term. For

10

example, when urbanization development level is considered as in medium stage,

11

 ( x) denotes

12

scenario.

13

(5) Similarity measure in referring to scenario five (S5)

middle ( x) . Model (4) is also applicable to the possibility (b) in this

14

In this scenario, both Z i and Z 0 assume an interval number for describing an

15

urban feature C j . In this circumstance, the feature C j for the reported city Z i is

16

ⅠⅡ ⅠⅡ described as pij  [ pij , pij ] , and that for the target city Z 0 is descried as p0 j  [ p0 j , p0 j ]

ACCEPTED MANUSCRIPT 1

.The similarity between Z i and Z 0 under this scenario, denoted by S 5i ( Z i , Z 0 ) , can be

2

obtained through the following formula (Slonim & Schneider, 2004): pⅡⅡ pij 0j

S 5（ i Z i , Z 0 )  1-

3

4

  p0 j

pⅠ ij

x  y dxdy

(    )( pⅡⅠⅡⅠ ij  pij )( p0 j  p0 j )

（5）

(6) Similarity measure in referring to scenario six (S6)

5

The two possibilities in this scenario include: (a) Z i assumes an interval number

6

and Z 0 assumes a fuzzy linguistic variable for describing the urban feature C j ; (b) Z i

7

assumes a fuzzy linguistic variable and Z 0 assumes an interval number.

8

In referring to the possibility (a), C j for the reported city Z i is specified as

9

Z 0 is a fuzzy linguistic variable p0 j , with the pij  [ pⅠⅡ ij , pij ] , and that for the target city

10

fuzzy membership function of  ( x) . In this case, according to Slonim & Schneider

11

(2004), the similarity between Z i and Z 0 in referring to scenario (6), can be obtained

12

from the follows: pⅡ ij

13

14 15



   ( x) x  p dxdy S 6 ( Z , Z )  1 (    )( p  p )   ( x)dx  ij

pⅠ ij

i

i

0

ⅡⅠ ij

ij

（6） The model (6) is also applicable to the possibility (b) of this scenario.

16 17

(7) Similarity measure in referring to scenario seven (S7)

18

In this scenario, the values of C j for both Z i and Z 0 are measured in fuzzy

19

linguistic terms. The fuzzy linguistic variables are pij and p0 j , with the membership

ACCEPTED MANUSCRIPT 1 2 3

function of  ( x) and  '( x) respectively. Referring to this scenario, according to Pal and Shiu (2004), the similarity between the reported city Z i and the target city Z 0 

S 7（） i Zi , Z0

4

follows: 

   ( x) '( y) x  y dxdy =1-     (    )   ( x)dx   '( y )dy  

(7)

5 6

(8) A hybrid similarity measure

7

The above formulas (1) - (7) are used in combination to determine the similarity

8

between Z i and Z 0 when individual urban feature C j (j=1, 2,…,6) is considered. To

9

retrieve the most similar city reported in experience database for good reference to a

10

given target city, the similarity values for measuring various urban features C j can be

11

integrated into a hybrid similarity measure. This integrated measurement will be used

12

to indicate the general similarity level between Z i and Z 0 , denoted as S ( Z 0 , Z i ) : 6

S (Z0 , Zi )   w j S j (Z0 , Zi )

13

(8)

j 1

14

Where

wj

is the weighting value for indicating the significance of the feature C j .

15

There are several methods introduced in previous studies for defining the weighting

16

values (Ha., 2008; Gançarski et al., 2008). S j ( Z 0 , Z i ) is the similarity value between

17

Z i and Z 0 from the perspective of the specific urban feature C j .

18 19 20

6. A demonstration of applying hybrid similarity measure This section demonstrates the application of the hybrid similarity measure

ACCEPTED MANUSCRIPT 1

developed in previous section. In applying the measure, a target city of Beijing was

2

selected, denoted as Z 0 , which has been facing the problem of increasing carbon

3

emission and smog. This problem is a global issue and cities around the world have

4

been taking various types of measures to address the problem. Ten international cities

5

are selected, which have gained experiences in mitigating carbon emissions, as reported

6

cities. The application of hybrid similarity measure is to help identify the most similar

7

city to the target city of Beijing. It is considered that the most similar city can offer the

8

effective experience or reference for assisting in solving the problem faced in Beijing.

9

(1) Data collection

10 11

The following ten best practices are collected from the website of New York City Global Partners to mitigate carbon emission:

12



Calgary City (Z1): Target to Use 100% Green Electricity

13



Copenhagen (Z2): Carbon Neutral by 2025

14



Mexico city (Z3): Comprehensive Climate Change Plan

15



Barcelona (Z4): Promoting Solar Energy

16



Berlin (Z5): Public-Private Partnership for Building Retrofits

17



Toronto (Z6): Deep Lake Water Cooling System

18



Tokyo (Z7): Green Building Program

19



Seoul (Z8): Eco-Mileage System

20



New York (Z9): NYC Greener, Greater Buildings Plan

21



Sao Paul (Z10): Landfill Emissions Control

ACCEPTED MANUSCRIPT 1

The backgrounds of these reported cities are described by six urban features

2

(C1 , C2 ,..., C6 ) . These data are collected from the website of Wikipedia, World Bank

3

and International Monetary Fund, as presented in Table 4.

4 5

(Insert Table 4 here)

6 7

(2) Data analysis

8

Urban features in Table 4 are described in four types of value formats, namely,

9

crisp symbol, crisp number, interval number, and fuzzy linguistic value. By applying

10

the data in Table 4 to the models (1)-(8), the results of similarity analysis are obtained,

11

as shown in Table 5.

12 13

(Insert Table 5 here)

14 15 16

Similarity in referring to the urban feature “landform (C1)”

17

The value of landform is measured by crisp symbol for all the ten reported cities

18

and the target city. The similarity analysis for the urban feature “landform” refers to the

19

scenario one (S1). Thus the model (1) is used to calculate the similarity between Beijing

20

and other ten cities. The calculation results are shown in the second column in Table 5.

21

For example, the similarity between Beijing and Mexico City in terms of landform is

ACCEPTED MANUSCRIPT 1

0.

2 3

Similarity in referring to the urban feature “climate (C2)”

4

The value of climate is measured in the format of the crisp symbol for Beijing and

5

ten reported cities. The similarity analysis in this case refers to the situation one (S1).

6

The analysis results are shown in the third column in Table 5. For example, the

7

similarity of climate between Beijing and Toronto is 1.

8

Similarity in referring to the urban feature of “population (C3)”

9

The values of population are measured in the format of crisp number for all case

10

cities, thus the similarity analysis in this case refers to the second scenario (S2).

11

Therefore, model (2) is used to calculate the similarity. The calculation results are

12

presented in the fourth column in Table 5. For example, the similarity of population

13

between Beijing and Tokyo is 0.731

14

Similarity in referring to the feature “urbanization level (C4)”

15

In Table 4, the value of urbanization level is described by fuzzy linguistic terms,

16

or crisp numbers, or interval numbers for various case cities. For example, the

17

urbanization level for Beijing is described in the format of the fuzzy linguistic term

18

“mature stage”, for which the fuzzy membership function is defined as

19

20

0  x  0.6  mature ( x)    0.8  0.6 1

0  x  0.6 0.6  x  0.8 0.8  x  1

Therefore, the similarity analysis for the feature “urbanization level” in this case

ACCEPTED MANUSCRIPT 1

refers to scenario four (S4) and scenario six (S6), and model (4) and model (6) are

2

therefore used. The calculation results are obtained accordingly, as shown in the fifth

3

column in Table 5. For example, the similarity between Beijing and Barcelona is 0.833

4

with regard to urbanization level.

5

Similarity in referring to the feature of “Gini coefficient (C5)”

6

In Table 4, Gini coefficient is described with fuzzy linguistic terms and crisp

7

number in case cities. For example, the Gini coefficient for the target city (Beijing) is

8

in the format of fuzzy linguistic term “Relative inequality”, and the Gini coefficient for

9

the reported city (Tokyo) is in the format of crisp number “0.381”. By referring to the

10

analysis method discussed in the Section 4, trapezoidal membership function is used to

11

describe fuzzy linguistic terms “Reasonable” and “Relative inequality”, and the

12

membership functions of the fuzzy linguistic variables are defined as: 0  x  0.25    R ( x)   0.1  0.45  x  0.1 0 

x  0.25 0.25  x  0.35 0.35  x  0.45 0.45  x

13 0  y  0.35    RI (y)   0.1  0.55  y  0.1 0 

14

y  0.35 0.35  y  0.45 0.45  y  0.55 0.55  y

The similarity analysis in this case refers to the scenario four (S4) and the scenario

ACCEPTED MANUSCRIPT 1

seven (S7), for which model (4) and model (7) are therefore adopted to produce

2

calculations. The results are presented in the sixth column in Table 5. For example, the

3

similarity of Gini coefficient between Beijing and Seoul is 0.96.

4

Similarity in referring to the urban feature of “GDP per capital (C6)”

5

It can be seen from Table 4 that the value of GDP per capital is measured in the

6

format of crisp number and interval number for case cities. Therefore, the similarity

7

analysis in this case refers to the scenario three (S3) and scenario five (S5), for which

8

model (3) and model (5) are therefore used. The calculation results are shown in the

9

seventh column in Table 5. For example, the similarity between Beijing and New York

10

is 0.085 from the perspective of GDP per capita.

11

Integrated similarity

12

To analyze the integrated similarity between Beijing and the ten reported cities by

13

applying model (8), weighting values need to be assigned to the six urban features. The

14

allocation of weighting values between the six urban features depends on the conditions

15

of specific problems. In this study, for the purpose of demonstration, the weighting

16

values

17

 W  (0.16, 0.17, 0.17, 0.17, 0.17, 0.16) . Accordingly, the integrated similarity values

18

are obtained, as shown in the last column in Table 4.

between

the

six

urban

features

are

evenly

distributed,

namely,

19 20

(3) Discussion

21

The demonstration suggests the good effectiveness of the proposed hybrid similarity

ACCEPTED MANUSCRIPT 1

measure in supporting city government to identify most useful experience in promoting

2

urban sustainability. By using the hybrid similarity measure in this case study, the

3

similarity level between Beijing and the ten reported cities have been analyzed. The

4

results show that individual reported cities have different integrated similarity levels

5

with Beijing, and Seoul has highest similarity with Beijing. This suggests that Seoul is

6

the most similar city to Beijing among the ten selected cities from the perspective of

7

six urban features. This result can be supported by the following discussions. Beijing is

8

geographically close to Seoul, and the two cities are nearly on the same latitude. They

9

have the same type of landform and climate. Both the two cities are national capital

10

cities, and they share extensive culture heritages. In line with this result, the experience

11

of Seoul is probably most useful among the ten reported cities to support the decision-

12

making on solution for mitigating carbon emissions problem in Beijing.

13

In further analysis on the best practice report of Seoul, it can be found that Seoul

14

has been implementing a citizen participation program called eco-mileage system in

15

2009. This system gives incentives to member households and organizations who

16

voluntarily reduce consumption of electricity, water, or gas use (NYC, 2015). In

17

applying this system Seoul government signed an agreement with manufacturers to

18

encourage them to provide high efficiency appliances. The Seoul government also

19

encouraged consumers to purchase the household environmental-friendly products and

20

incentive energy saving behavior of individuals by monitoring household’s energy

21

consumption. Furthermore, a collaboration of city administration, financial institutions,

ACCEPTED MANUSCRIPT 1

manufacturers, and retailers was established to lighten the burden of the city

2

government. Beijing’s government has also proposed the similar energy saving policy

3

to reduce carbon emissions in November 2015 (Beijing Municipal Commission of

4

Commerce, 2015), which encourages consumers to buy environmental-friendly

5

products by giving subsidies. These facts are echoed by the results generated from our

6

hybrid similarity analysis. Therefore, the eco-mileage system adopted in Seoul is

7

recommended as good experience for Beijing government to improve its energy-saving

8

policy. Best practices proposed by other cities could be references in formulating

9

solutions for Beijing government to reduce carbon emissions. However, their solutions

10

may not be most suitable for Beijing. For example, Calgary City and Copenhagen all

11

adopted wind power to generate electricity which could not be promoted widely in

12

Beijing due to the limitation of climate. Hence, the experiences in fight against emission

13

in Seoul have better value for Beijing to mitigate emission than that from other reported

14

cities.

15 16

6. Conclusion

17

In the process of promoting sustainable urbanization, many cities around the world

18

have devoted great efforts and presented a large number of successful practice cases.

19

These successful practices provide valuable experiences which are potentially useful

20

for assisting other cities especially in those underdeveloped regions to tackle challenges

21

in urbanization practice. However, government decision makers often adopt the

ACCEPTED MANUSCRIPT 1

existing experiences indiscriminately without the consideration of the background

2

difference between the target city and those cities which have successful experiences.

3

It is an unaddressed issue for government decision makers which city should be selected

4

as the experience-provider. This limitation exists in the Experience-Mining System

5

(ExMS) introduced by Shen et al. (2013) in exploring experience of sustainable

6

urbanization. This study introduces an innovative method for mitigating this significant

7

limitations. The innovative method is characterized by introducing the concept of urban

8

similarity and a hybrid similarity measure. In the hybrid measure, six urban feature

9

variables are included which assume four types of value formats. The presence of

10

different value formats leads to the formulation of seven scenarios in the process of

11

analyzing the measurement. The application of this hybrid measure method makes it

12

possible to share effectively the good experiences generated under similar

13

circumstances in applying ExMS. The similarity measure mitigates the errors caused

14

by taking experiences indiscriminately from other practices.

15

This study provides a new perspective to address urban sustainability problems by

16

introducing urban similarity. The innovative method proposed contributes to the

17

development of current theoretical framework in the field of experience mining for

18

decision making. From the perspective of practices, the outcome of this study can assist

19

government policy makers to select valuable experiences from existing best practices

20

by considering the similarity of urban features, thus the selected experiences will be

21

effective in promoting the development of sustainable urbanization. Furthermore, the

ACCEPTED MANUSCRIPT 1

theoretical framework of this study can be extended to other domains, for example the

2

application of green building practices.

3

The limitation of this study is appreciated. The number of urban features is limited

4

in describing a city, although it is very difficult to describe a city in a perfect way due

5

to the complexity of the city itself. It is recommended for future research that more

6

urban features should be added to improve the effectiveness in using the hybrid

7

measure.

8 9 10

Acknowledgments

11

The authors would like to acknowledge the financial support for this research

12

received from the National Social Science Foundation of China (Grant No.

13

15AZD025).

14 15

Reference

16

Atkinson, A. B., 2002. Social indicators: The EU and social inclusion. Oxford University

17 18

Press. Beijing Municipal Commission of Commerce, 2015. The announcement of implementing

19

energy-saving

20

http://beijing.mofcom.gov.cn/article/sjtongzhigg/201511/20151101195492.shtml

21

and

emission

reduction

policy.

Retrieved

from:

Behbahani, M., Saghaee, A., Noorossana, R., 2012. A case-based reasoning system

ACCEPTED MANUSCRIPT 1

development for statistical process control: Case representation and retrieval. Computers &

2

Industrial Engineering, 63(4), 1107-1117.

3

Boschmann, E. E., & Gabriel, J. N. (2013). Urban sustainability and the LEED rating system:

4

case studies on the role of regional characteristics and adaptive reuse in green building in Denver

5

and Boulder, Colorado. The Geographical Journal, 179(3), 221-233.

6

Carmona, M. A., Barbancho, J., Larios, D. F., León, C., 2013. Applying case based reasoning

7

for prioritizing areas of business management. Expert Systems with Applications, 40(9), 3450-3458.

8

Castro, J. L., Navarro, M., Sánchez, J. M., Zurita, J. M., 2009. Loss and gain functions for CBR

9

retrieval. Information Sciences, 179(11), 1738-1750.

10

Chang, P. C., Hieh, J. C., Liao, T. W., 2005. Evolving fuzzy rules for due-date assignment

11

problem in semiconductor manufacturing factory. Journal of Intelligent Manufacturing, 16(4-5),

12

549-557.

13

Chelleri, L., Schuetze, T., & Salvati, L. (2015). Integrating resilience with urban sustainability

14

in neglected neighborhoods: Challenges and opportunities of transitioning to decentralized water

15

management in Mexico City. Habitat International, 48, 122-130.

16 17

Chen, D., & Chen, H. W., 2013. Using the Köppen classification to quantify climate variation and change: an example for 1901–2010. Environmental Development, 6, 69-79.

18

Dedekorkut-Howes,

19

playground?. Cities, 42, 70-84.

20 21

A.,

&

Bosman,

C.,

2015.

The

Gold

Coast:

Australia’s

Donald, J. (1992). Metropolis: the city as text. Social and cultural forms of modernity, 417470.

ACCEPTED MANUSCRIPT 1

Dye, C., 2008. Health and urban living. Science, 319(5864), 766-9.

2

Dyson, T., 2011. The role of the demographic transition in the process of urbanization.

3

Population & Development Review, 37(Suppl 1), 34-54.

4

Economic development council of Small and Medium-sized Cities in China Society of Urban

5

Economy., 2010. Annual Report on Development of Small and Medium-Sized Cities in China.

6

Beijing: Social Sciences Academic Press.

7

El-Fakdi, A., Gamero, F., Melendez, J., Auffret, V., Haigron, P. (2014). eXiTCDSS: A

8

framework for a workflow-based CBR for interventional Clinical Decision Support Systems and its

9

application to TAVI. Expert Systems with Applications, 41(2), 284-294.

10

Ernst, L., Dinther, G. V., Peek, G. J., & Loorbach, D. A., 2016. Sustainable urban

11

transformation and sustainability transitions; conceptual framework and case study. Journal of

12

Cleaner Production, 112, 2988-2999.

13 14

European Commission. (2006). Targeted summary of the European sustainable cities report for local authorities. European Commission.

15

Fan, Z. P., Li, Y. H., Wang, X., & Liu, Y., 2014. Hybrid similarity measure for case retrieval

16

in CBR and its application to emergency response towards gas explosion. Expert Systems with

17

Applications, 41(5), 2526-2534.

18 19 20 21

Fan, Z. P., Li, Y. H., & Zhang, Y., 2015. Generating project risk response strategies based on CBR: A case study. Expert Systems with Applications, 42(6), 2870-2883. Flynn, A., Yu, L., Feindt, P., & Chen, C. (2016). Eco-cities, governance and sustainable lifestyles: the case of the Sino-Singapore Tianjin eco-city. Habitat International, 53, 78-86.

ACCEPTED MANUSCRIPT 1 2 3 4 5 6

Gan, X., Zuo, J., Ye, K., Skitmore, M., & Xiong, B. (2015). Why sustainable construction? Why not? An owner's perspective. Habitat International, 47, 61-68. Gan, X., Zuo, J., Ye, K., Li, D., Chang, R., & Zillante, G. (2016). Are migrant workers satisfied with public rental housing? A study in Chongqing, china. Habitat International, 56, 96-102. Gançarski, P., Blansche, A., & Wania, A., 2008. Comparison between two coevolutionary feature weighting algorithms in clustering. Pattern Recognition, 41(3), 983-994.

7

Gao, H., 1995. Western economics. Beijing: Economy Press China.

8

Castro, J. L., Navarro, M., Sánchez, J. M., Zurita, J. M., 2009. Loss and gain functions for CBR

9

retrieval. Information Sciences, 179(11), 1738-1750.

10

Guessoum, S., Laskri, M. T., & Lieber, J., 2014. RespiDiag: a case-based reasoning system for

11

the diagnosis of chronic obstructive pulmonary disease. Expert Systems with Applications, 41(2),

12

267-273.

13 14 15 16 17 18 19 20 21

Ha, S., 2008. A personalized counseling system using case-based reasoning with neural symbolic feature weighting (CANSY). Applied Intelligence, 29(3), 279-288. Hjaltason, G. R., & Samet, H., 2003. Index-driven similarity search in metric spaces (survey article). ACM Transactions on Database Systems (TODS),28(4), 517-580. International Monetary Fund, 2015. Data Mapper, World Economic outlook. Retrieved from: http://www.imf.org/external/datamapper/index.php. Intergovernmental Panel on Climate Change. (2015). Climate change 2014: mitigation of climate change (Vol. 3). Cambridge University Press. Khanna, N., Fridley, D., & Hong, L. (2014). China's pilot low-carbon city initiative: A

ACCEPTED MANUSCRIPT 1

comparative assessment of national goals and local plans. Sustainable Cities and Society, 12, 110-

2

121.

3 4 5

Köppen, W., 1900. Versuch einer Klassifikation der Klimate, vorzugsweise nach ihren Beziehungen zur Pflanzenwelt. Geographische Zeitschrift, 6(11), 593-611. Kong, W., Chai, T., Yang, S., & Ding, J., 2013. A hybrid evolutionary multiobjective

6

optimization

7

manufacturing. Applied Soft Computing, 13(5), 2960-2969.

8 9

strategy

for

the

dynamic

power

supply

problem

in

magnesia

grain

Krajnc, D., & Glavič, P., 2005. A model for integrated assessment of sustainable development. Resources, Conservation and Recycling, 43(2), 189-208.

10

Kuo, T. C., 2010. Combination of case-based reasoning and analytical hierarchy process for

11

providing intelligent decision support for product recycling strategies. Expert Systems with

12

Applications, 37(8), 5558-5563.

13

Landry, C., 2006. The art of city-making. Routledge.

14

Li, Y., & Qiu, L. (2015). A comparative study on the quality of China's eco-city: Suzhou vs

15 16 17 18 19 20 21

Kitakyushu. Habitat International, 50, 57-64. Liao, T. W., 2004. An investigation of a hybrid CBR method for failure mechanisms identification. Engineering Applications of Artificial Intelligence,17(1), 123-134. Liao, T. W., Zhang, Z., Mount, C. R., 1998. Similarity measures for retrieval in case-based reasoning systems. Applied Artificial Intelligence,12(4), 267-288. Liu, H., Zhou, G., Wennersten, R., & Frostell, B. (2014). Analysis of sustainable urban development approaches in china. Habitat International, 41(1), 24–32.

ACCEPTED MANUSCRIPT 1 2

Liu, Q., & Xi, J., 2011. Case-based parametric design system for test turntable. Expert Systems with Applications, 38(6), 6508-6516.

3

Liu, Z., Adams, M., Cote, R. P., Geng, Y., & Li, Y., 2016. Comparative study on the pathways

4

of industrial parks towards sustainable development between China and Canada. Resources,

5

Conservation and Recycling.

6 7 8 9

Mccormick, K., Neij, L., Anderberg, S., Coenen, L., 2013. Advancing sustainable urban transformation. Journal of Cleaner Production, 50(13), 1-11. Meng, M., Koh, P. P., Wong, Y. D., & Zhong, Y. H., 2014. Influences of urban characteristics on cycling: Experiences of four cities. Sustainable Cities and Society, 13, 78-88.

10

Moussiopoulos, N., Achillas, C., Vlachokostas, C., Spyridi, D., & Nikolaou, K., 2010.

11

Environmental, social and economic information management for the evaluation of sustainability

12

in urban areas: A system of indicators for Thessaloniki, Greece. Cities, 27(5), 377-384.

13

Mutisys, E., Yarime, M., 2014. Moving towards urban sustainability in Kenya: a framework

14

for integration of environmental, economic, social and governance dimensions. Sustainability

15

Science, 9, 205-215.

16 17 18 19 20 21

National

Geographic,

2015.

Education,

Landform.

Retrieved

from:

http://education.nationalgeographic.com/education/encyclopedia/landform/?ar_a=1 NYC, 2015. New York City Global Partners-Innovation Exchange, New York City Government. Retrieved from http://www.nyc.gov/html/ia/gprb/html/search/search.shtml NYC,

2015.

Eco-Mileage

System.

Retrieved

http://www1.nyc.gov/site/globalpartners/programs/search-best-practice-reports.page

from:

ACCEPTED MANUSCRIPT 1

Northam, R. M., 1979. Urban geography. John Wiley & Sons.

2

Pal, S. K., & Shiu, S. C., 2004. Foundations of soft case-based reasoning(Vol. 8). John Wiley

3

& Sons.

4 5

Pereira, I., & Madureira, A., 2013. Self-optimization module for scheduling using case-based reasoning. Applied Soft Computing, 13(3), 1419-1432.

6

Perner, P., 2014. Mining Sparse and Big Data by Case-based Reasoning.Procedia Computer

7

Science, 35, 19-33.Russo, F., Comi, A., 2012. City characteristics and urban goods movements: A

8

way to environmental transportation system in a sustainable city. Procedia-Social and Behavioral

9

Sciences, 39, 61-73.

10 11

Schmidt, S., Magigi, W., & Godfrey, B., 2015. The organization of urban agriculture: Farmer associations and urbanization in Tanzania. Cities, 42, 153-159.

12 13

Shen, L., Shuai, C., Jiao, L., Tan, Y., & Song, X. (2017). Dynamic sustainability performance during urbanization process between BRICS countries. Habitat International, 60, 19-33.

14

Shen, L. Y., Ochoa, J. J., Zhang, X., & Yi, P., 2013. Experience mining for decision making

15

on

16

Construction, 29, 40-49.

17 18 19 20 21

implementing

sustainable

urbanization—An

innovative

approach. Automation

in

Shen, L., Shuai, C., Jiao, L., Tan, Y., & Song, X. (2016). A global perspective on the sustainable performance of urbanization. Sustainability, 8(8), 783. Shi, Q., Lai, X., Xie, X., & Zuo, J. (2014). Assessment of green building policies–A fuzzy impact matrix approach. Renewable and Sustainable Energy Reviews, 36, 203-211. Shuai, C., Shen, L., Jiao, L., Wu, Y., & Tan, Y. (2017). Identifying key impact factors on

ACCEPTED MANUSCRIPT 1

carbon emission: Evidences from panel and time-series data of 125 countries from 1990 to 2011.

2

Applied Energy, 187, 310-325.

3 4

Slonim, T. Y., & Schneider, M., 2001. Design issues in fuzzy case-based reasoning. Fuzzy Sets and Systems, 117(2), 251-267.

5

Tadić, S., Zečević, S., & Krstić, M. (2014). A novel hybrid MCDM model based on fuzzy

6

DEMATEL, fuzzy ANP and fuzzy VIKOR for city logistics concept selection. Expert Systems with

7

Applications, 41(18), 8112-8128.

8 9 10 11 12 13 14 15 16

The

World

Bank,

2015.

Indicators,

GINI

index.

Retrieved

from:

http://data.worldbank.org/indicator/SI.POV.GINI. United Nations Human Settlements Programme, 2015. Best Practices Database in Improving the Living Environment. Retrieved from: http://mirror.unhabitat.org/bp/bp.list.aspx/. United Nations. Dept. of Economic., 2001. Indicators of sustainable development: Guidelines and methodologies. United Nations Publications. United Nations. Department of Economic and Social Affairs. Population Division., 2015. World urbanization prospects: The 2014 revision. UN. United Nations, Department of Economic and Social Affairs, 2015. Millennium Development

17

Goals

18

http://mdgs.un.org/unsd/mdg/Host.aspx?Content=Indicators%2fOfficialList.htm

Indicators.

Retrieved

from:

19

Vilsmaier, U., Engbers, M., Luthardt, P., Maas-Deipenbrock, R. M., Wunderlich, S., & Scholz,

20

R. W., 2015. Case-based Mutual Learning Sessions: knowledge integration and transfer in

21

transdisciplinary processes. Sustainability Science, 10(4), 563-580.

ACCEPTED MANUSCRIPT 1

Wang, Y., Han, Q., Vries, B. D., & Zuo, J. (2016). How the public reacts to social impacts in

2

construction projects? A structural equation modeling study. International Journal of Project

3

Management, 34(8), 1433-1448.

4

Wu, H., Wang, J., Duan, H., Ouyang, L., Huang, W., & Zuo, J. (2016). An innovative approach

5

to managing demolition waste via GIS (geographic information system): a case study in Shenzhen

6

city, China. Journal of Cleaner Production, 112, 494-503.

7 8 9 10 11 12 13 14

Xiong, N., 2011. Learning fuzzy rules for similarity assessment in case-based reasoning. Expert systems with applications, 38(9), 10780-10786. Yang B., Xu T., Shi L., 2017. Analysis on sustainable urban development levels and trends in China's cities. Journal of Cleaner Production, 141, 868-880. Yu, L. (2014). Low carbon eco-city: new approach for Chinese urbanisation. Habitat International, 44, 102–110. Zarandi, M. F., Razaee, Z. S., & Karbasian, M., 2011. A fuzzy case based reasoning approach to value engineering. Expert Systems with Applications,38(8), 9334-9339.

15

Zhai, X., Reed, R., & Mills, A., 2014. Embracing off-site innovation in construction in China

16

to enhance a sustainable built environment in urban housing. International Journal of Construction

17

Management, 14(3), 123-133.

18

Zhan, Y., Tan, K. H., Ji, G., Chung, L., & Chiu, A. S., 2016. Green and lean sustainable

19

development path in China: Guanxi, practices and performance. Resources, Conservation and

20

Recycling.

21

Zhang, X., 2016. Sustainable urbanization: a bi-dimensional matrix model. Journal of Cleaner

ACCEPTED MANUSCRIPT 1

Production, 134, 425-433.

2

Zhao, J., Deng, W., Song, Y., 2014. Ridership and effectiveness of bike sharing: The effects

3

of urban features and system characteristics on daily use and turnover rate of public bikes in

4

China. Transport Policy, 35, 253-264.

5 6 7 8

Zuo, J., & Zhao, Z. Y. (2014). Green building research–current status and future agenda: A review. Renewable and Sustainable Energy Reviews, 30, 271-281.

ACCEPTED MANUSCRIPT

Reported city (or Target City) Crisp symbol Crisp number Interval number

T a r g e t city ( o r Reported City) S1 S2 S3 S4 S5

Crisp symbol Crisp number Interval numbers

S6 Fuzzy linguistic value

S7

Fuzzy linguistic values

Similarity Figure 1 Seven scenarios of similarity between reported city and target city

ACCEPTED MANUSCRIPT

 ( x) 1

0

middle

early

0.2

0.4

0.6

mature

0.8

Figure 2 Fuzzy membership function

x (Urbanization level)

ACCEPTED MANUSCRIPT

Table 1 Urban features and their indicator/attribute Urban feature

Indicator/Attribute

Landform (C1)

Hills, Mountains, Plains, Plateaus

Climate (C2)

Tropical, Dry, Mild temperate, Snow, Polar

Urban scale (C3)

Total population

Development stage (C4)

Urbanization rate

Income distribution (C5)

Gini Coefficient

Economic performance (C6)

GDP per capita

ACCEPTED MANUSCRIPT

Table 2 Description of landform indicators Attributes Hills

Description Land that rises above its surrounding and has a rounded summit, usually less than 300 meters.

Mountains

Landmass that forms as tectonic plates interact with each other.

Plateaus

Large region that is relatively flat and higher than the surrounding area.

Plains

A plain is a broad area of relatively flat land at a low elevation.

Source: National Geographic, 2014

ACCEPTED MANUSCRIPT

Table 3 Köppen climate classification system Major group

Sub-types

Tropical

Tropical rain forest; Tropical monsoon; Tropical wet and dry savanna

Dry

Desert (arid); Steppe (semi-arid)

Mild temperate

Mediterranean; Humid subtropical; Oceanic

Snow

Humid; Subarctic

Polar

Tundra; Ice cap

Source: Chen & Chen, 2013

Table 4 Urban features of ten reported cities and the target city Climate (C2)

City

Landform (C1)

Calgary(Z1)

Plateau

Copenhagen (Z2) Mexico City (Z3)

Plain Plateau

Barcelona(Z4) Berlin(Z5) Toronto(Z6)

Plain Plain Plain

Tokyo(Z7)

Plain

Seoul(Z8)

Plain

SÃO PAULO(Z9)

Plateau

New York(Z10)

Plain

Beijing(Z0)

Plain

Humid continental climate Oceanic climate zone Subtropical highland climate Mediterranean climate Oceanic climate Humid continental climate Humid subtropical climate Humid continental climate Humid subtropical climate Humid subtropical climate Semi-humid continental monsoon climate

Urban feature

Population (C3)

Urbanization Level (%) (C4)

Gini index (%) (C5)

GDP per capita ($) (C6)

1,096,833

81.5

32.6

61586

569,557 8,851,080

87.3 78.7

24.0 48.1

39291 [19888,21135]

1,620,943 3,517,424 2,615,060

79.1 74.9 81.5

34.7 28.3 32.6

36280 33311 43905

13,185,502

[86,92.5]

38.1

41446

10,117,909

82.2

Reasonable

32155

11,895,893

85.2

54.7

23704

8,336,697

81.3

45.0

[62923,66727]

17,837,000

Mature stage

Relative inequality

[15000,20000]

Table 5 Similarity between Beijing and ten reported cities City Calgary Copenhagen Mexico City Barcelona Berlin Toronto Tokyo Seoul SÃO PAULO New York

S1 ( Z i , Z 0 )

S2 ( Zi , Z 0 )

S3 ( Z i , Z 0 )

S4 ( Zi , Z 0 )

S5 ( Z i , Z 0 )

S6 ( Z i , Z 0 )

S (Z0 , Zi )

0 1 0 1 1 1 1 1 0 1

1 0 0 0 0 1 0 1 0 0

0.031 0.000 0.480 0.061 0.171 0.119 0.731 0.553 0.656 0.450

0.832 0.807 0.832 0.833 0.809 0.832 0.841 0.788 0.820 0.832

0.752 0.503 0.916 0.794 0.666 0.752 0.86 0.96 0.805 0.933

0.147 0.579 0.942 0.637 0.694 0.489 0.537 0.716 0.88 0.085

0.468 0.475 0.529 0.549 0.551 0.698 0.659 0.836 0.529 0.550