Empowering Urban Public Transport Planning Process for Medium-Sized Cities in Developing Countries: Innovative Decision Support Framework for Sustainability

Abstract

The challenges resulting from rapid economic growth, urbanization, and increased motorization in developing nations necessitate a comprehensive and sustainable approach to urban public transport planning. While sustainable urban public transport (SUPT) planning offers a solution, the complexity of choosing suitable policy measure options remains a challenge. This study first introduces a decision support framework (DSF) that integrates the sustainable urban public transport manual (SUPTM) adopted for generating the potential SUPT policy measure options, the KonSULT knowledge base applied for providing the performance scores of each measure option for all determined criteria, and the HMADM (including FAHP, FSM, and TOPSIS) technique to create, rank, and select SUPT policy measure options tailored to medium-sized urban areas in developing nations. A case study of Khon Kaen City, Thailand, illustrates the practical application of the framework, resulting in a set of 31 (91.2%) out of the total of 34 ranked policy measure options. Comparing these prioritizations with the city’s existing plan reveals a substantial agreement, which suggests the potential applicability of the DSF. Overall, the DSF marks a significant advancement in SUPT planning, which is crucial for shaping efficient, equitable, and environmentally conscious urban mobility in developing countries, which are undergoing transformative change.

1. Introduction

Developing countries are transforming owing to rapid economic growth, population expansion, urbanization, and increased motorization [1]. Generally, urban areas are much more populated than rural areas. In 2018, 55% of the global population lived in urban areas. According to estimations, by 2050, 68% of the global population will be living in urban areas [2]. In 2022, Asian countries had 4.4 billion people (55% of the global urban population) [3]. The rapid urbanization of Asia has introduced challenges such as urban sprawl, car dependency, insufficient transport infrastructure, and inefficient urban public transport systems [2,4]. This crisis has led to problems such as traffic congestion, accidents, environmental damage, inefficient energy use, social inequality, global warming, and climate change.

Approximately 50% of the global population lives in urban cities (with less than 500,000 residents) [2]. Medium-sized cities in developing countries are typically home to urban residents (between 200,000 and 500,000 people) who face many transport challenges [2,5]. While such cities can accommodate urban expansion, facilitate sustainable modes of transportation (e.g., public transit and active transport), and implement environmental preservation measures, these cities lack the resources to implement any sustainable urban public transport (SUPT) policy measures. Moreover, the inefficiencies in urban public transport systems and the high energy consumption per capita in these cities are urgent concerns, both of which are influenced by their spatial dimensions and population densities [6]. Addressing these complicated challenges requires SUPT planning.

Since 2013, EU nations have created and executed sustainable urban mobility plans (SUMPs) [7]. SUMP documents offer valuable strategic resources [8], including CH4LLENGE [9], CIVITAS SUMPs-UP [10], and EVIDENCE [11]. Based on the CH4LLENGE principle, a powerful web-based knowledge base system known as the knowledgebase on sustainable urban land use and transport (KonSULT) [12] was initially developed to support decision makers, urban transport planners, and engineers, as well as personnel from related sectors in creating, ranking, and choosing suitable sustainable urban land use and transport (SULT) policy measures as well as their packages. This platform provides contribution scores for these measure options, in keeping with all seven objectives. These scores were achieved from the professional experiences and expertise of empirical studies and human experts in various EU countries [13,14,15,16,17].

The generation, prioritization, and selection of SUPT policy options have been rarely regarded as vital in the SUMP process in developed countries or developing nations [13]. Ideally, such a process should be applicable to cities in both developed and developing nations [13]. However, developed and developing cities vary significantly in area contexts (e.g., sizes and growth rates of population, urbanization, and motorization, etc.), challenges encountered (e.g., congestion, road accidents, environmental impacts, climate change, and lack of resources, etc.), existing transportation systems (e.g., inefficient and insufficient public transport systems, etc.), and the availability of policy options (e.g., only low-cost public transport systems and paratransit systems, etc.), among others [13,18,19]. May et al. [13] noted that current KonSULT software may not be directly applicable to developing cities. Therefore, further investigations of its potential, practicality, applicability, and viability are crucial for developing systematic and applicable procedures for creating, ranking, and proposing suitable policy measures for SUPT planning in developing nations.

The generation of the SUMP and, therefore, SUPT options is complicated and difficult for the following reasons: (1) an excessive emphasis on predetermined ideas in choosing economically feasible and acceptable options, (2) a lack of evidence of the performances of each option for achieving all defined objectives, and (3) a lack of awareness regarding the availability of diverse options [13,19]. In this study, all 34 policy measure options available in the national sustainable urban public transport manual (SUPTM) of Thailand [20] were adopted. The performance score of each determined option according to all seven minor criteria (objectives) has never been developed in developing nations, including Thailand. Therefore, the KonSULT knowledge base [12] was applied to assign the contribution scores of each policy measure option in association with the seven determined objectives.

The hybrid multi-attribute decision-making (HMADM) approach involving the fusion of several straightforward and beneficial multi-attribute decision-making (MADM) approaches has the potential for more accurate outcomes, albeit at the cost of increased complexity and difficulty [21,22,23]. Prioritizing and selecting policy measures are complex and characterized by ambiguity, uncertainty, and the lack of a unique structure. This decision-making challenge involves multiple criteria (both tangible and intangible), contentious group judgments, a definite number of options, and the involvement of uncertain, incomplete, and ambiguous information. The HMADM method aligns well with the inherent nature of such decision problems, making it suitable for addressing this complexity [24,25]. A novel HMADM approach integrating fuzzy analytic hierarchy process (FAHP) [26], fuzzy scoring method (FSM) [27], and technique for order of preference by similarity to ideal solution (TOPSIS) [28,29] is utilized in this study. The FAHP, FSM, and TOPSIS are adopted for their individual robustness in terms of theoretical foundations, advantages, and applicability. The FAHP is used to determine the relative weights of criteria in uncertain or ambiguous environments [30,31]. The FAHP exhibits accuracy and rationality through its hierarchical structure, which ensures the clarity and transparency of decision elements, as well as the direct measurement of expert judgment consistency. The FSM is applied to transform uncertain and incomplete (fuzzy) information into precise numerical data using straightforward calculation steps. In contrast, the TOPSIS is used to estimate the composite score of each option because of its rigor, logic, simplicity, transparency, and popularity.

Research on creating, ranking, and selecting the policy measure options for SUPT planning in medium-sized cities in developing countries is lacking. To fill the gap, this study aims to develop a decision support framework (DSF) that integrates the national SUPTM manual of Thailand, the robust KonSULT knowledge base (KB), with new HMADM techniques, including FAHP, FSM, and TOPSIS. The DSF is used to generate, prioritize, and choose appropriate SUPT policy options for medium-sized urban cities in developing countries. Khon Kaen City (KKC), Thailand, was adopted as a representative medium-sized city. The following is proposed as the research objective: “Can the newly proposed DSF be effectively applied to the SUPT planning process in any medium-sized city in developing countries”?

The remainder of this paper is organized as follows: Section 1 introduces the challenge of combining a SUPT planning process and the HMADM approach in developing nations. Section 2 presents a concise overview of the relevant literature. Section 3 describes the framework employed in the DSF assessment. Section 4 presents the results of the analysis. Section 5 presents the discussion. Finally, Section 6 presents the conclusions.

2. Literature Review

2.1. SUPTM Application

In 2015, the Office of Transport and Traffic Policy and Planning (OTP), Ministry of Transport, Thailand, developed the national SUPTM of Thailand to guide decision makers, administrators, transport engineers, and planners in the SUPT planning process for 78 regional cities across Thailand. The SUPTM manual defines the vision of SUPT planning for KKC as “developing standardized urban public transport systems that are convenient, comfortable, fast, reliable, easily accessible and connected, economical, safe, efficient, environmentally friendly, and sustainable” [20]. The seven key objectives of SUPT planning [20,32] are defined as follows: (1) economic efficiency; (2) economic growth; (3) finance; (4) livable streets and neighborhoods; (5) equity and social inclusion; (6) safety; and (7) environmental protection. Based on the SUPTM manual [20], the three principal strategies are as follows: (1) Strategy A (avoid: avoiding or reducing traveling by car, but emphasizing walking and cycling and applying urban land use development planning); (2) Strategy B (better: improving the efficiency of the existing public transport system, including the network of public transport service routes); and (3) Strategy C (change: changing travel patterns from private vehicles to efficient public transport systems, among others). A total of 34 policy measures were proposed in the national SUPTM manual, and these policy measures were classified into five groups: (1) design and planning, (2) land use and transport infrastructure, (3) regulations and management, (4) communications and technologies, and (5) economy and subsidy [20].

2.2. KonSULT KB Application

To facilitate effective integration in urban mobility plans, the Procedures for Recommending Optimal Sustainable Planning of European City Transport Systems (PROSPECTS) project [33] was used in the development of the Decision Maker’s Guidebook (DMG) [34]. This guidebook was designed to assist decision makers in formulating SULT plans. A crucial element within the logical framework proposed in the DMG is the process of generating, ranking, and choosing policy measure options for various types of urban areas [34].

In KonSULT, 64 policy measures for SULT development are categorized into six groups: land use, infrastructure, management and service, attitudes and behavior, information provision, and pricing [12]. May [9] used KonSULT KB to devise a decision support tool to create potential policy packages. Subsequently, the effectiveness of KonSULT software in formulating and selecting a SULT policy was assessed [13].

KonSULT is useful for strategizing SULT planning and development in urban areas by generating, ranking, and selecting specific policy measures and their packaging [8,9,12,13]. KonSULT employs contribution scores to evaluate the alignment of each policy measure with an individual objective (problem or indicator). These scores are assigned on an 11-point scale (ranging from −5 (strongest possible negative contribution) to +5 (strongest possible positive contribution)) and covering both positive and negative values. The contribution scores were generally obtained from human knowledge and expertise from professional experience on SULT planning and development in European countries.

2.3. Hybrid Multiple-Attribute Decision Making (HMADM)

MADM is a widely used approach for prioritizing, evaluating, and selecting the most suitable options from a defined set of discrete choices. This process considers multiple decision criteria and the relative importance assigned to these criteria. MADM techniques are crucial and popular for addressing practical, real-world challenges involving multiple decision criteria, goals, and potentially conflicting objectives [35]. Various well-recognized MADM methods have been developed. Each technique has distinct strengths, weaknesses, applicability, and restrictions [36].

Additionally, the HMADM approach, which combines several robust and applicable MADM algorithms, offers more precise and beneficial outcomes, despite the trade-off of increased computational complexity [21]. In practice, decision making often involves uncertain, incomplete, and ambiguous (fuzzy) information. Failing to incorporate such fuzzy information could potentially limit the effectiveness and capabilities of HMADM.

In HMADM studies [30,37,38], the analytic hierarchy process (AHP) is commonly used to determine the relative weights of each criterion. AHP-TOPSIS has emerged as a notable HMADM methodology, and it has been applied by several researchers [36]. AHP, TOPSIS, and PROMETHEE are among the most frequently utilized MADM techniques owing to their versatility, transparency, robust algorithms, and software availability [36].

Table 1 provides an overview of past applications of various HMADM techniques for SULT and SUPT ranking, evaluation, and selection. This table illustrates the range of approaches previously employed in tackling various issues within SULT and SUPT planning processes. Specifically, numerous MADM techniques have been developed and embraced to tackle multicriteria decision-making challenges [24,25,39,40]. May et al. [17] underscore the deficiency in transport planning regarding limited option generation, leading to suboptimal policies. They introduce KonSULT, a decision support tool, which identifies synergistic combinations of policy instruments to generate effective policy packages. The tool’s scoring system, initially based on professional judgment and later refined with a sketch planning model, highlights the best-performing instruments, facilitating improved policy formulation and implementation. Hamurcu and Eren [41] prioritized public transportation projects in Kırıkkale, Turkey, with a focus on enhancing urban livability. They used AHP and TOPSIS methods. Damidavičius et al. [42] aimed to develop and implement mobility measures for sustainable transport in Lithuania. They utilized various methods like COPRAS, TOPSIS, ARAS, EDAS, and WAM. Hamurcu and Eren [43] selected electric buses for sustainable transportation in Turkey using multi-criteria decision analysis methods, such as AHP and TOPSIS. Goyal et al. [35] investigated performance measures and proposed recommendations for sustainable transportation development in India, employing TOPSIS, VIKOR, ÉLECTRE, Delphi, and FAHP methods. Lastly, Mesa et al. [44] prioritized policy measure options for sustainable urban land use and transport development in KKU, Thailand, using AHP and TOPSIS methods. This concise literature review of both theoretical and empirical studies reveals that AHP, FAHP, and TOPSIS are the most prominent MADM methods in these sectors.

The key challenge is choosing the most appropriate MADM combination to formulate the HMADM approach for this study. The FAHP, FSM, and TOPSIS methods were selected due to the inherent robustness of the theoretical algorithms and their application potential. The FAHP has been mainly utilized to determine the relative weights of various criteria in fuzzy environments [30,31]. The FAHP achieves simplicity, adaptability, accuracy, and applicability using a hierarchical structure to ensure the rationality and transparency of decision criteria. It adopts the pairwise comparisons via the 9-point ratio scale system to achieve higher preciseness than traditional absolute scoring techniques in gauging the expert judgment consistency, dealing with the group judgments, and considering the inherent fuzzy information and judgments [24,45,46,47,48,49,50]. The FSM is used to convert linguistic (fuzzy) scores into precise numerical (crisp) scores because its theoretical foundation is rigorous, reasonable, accurate, and applicable; its computation steps are straightforward; and its historical applications are successful and acceptable [24,28,29,47,48,49,50]. TOPSIS is used to estimate the composite score of each option due to its theoretical algorithm based on the ideal point and Euclidean distance principle, which is robust, logical, transparent, accurate, and adaptable. TOPSIS has been successfully applied in a wide variety of research and practical fields [24,39,47,48,49,50,51].

Table 1. HMADM applications for ranking and evaluating the SULT and SUPT sectors.

Objectives	Applications	Applied *	Fuzzy Environment	Area (Year)	References
To introduce KonSULT, a decision support tool, which identifies synergistic combinations of policy instruments to generate effective policy packages.	Generating and ranking of SULT policy measure	Weighting and scoring methods	-	Europe (2012)	[17]
To introduce a decision support procedure for analyzing and generating consensus among various stakeholders in a transit development.	Utilization of the decision support procedure	FAHP and interval AHP	✓	Turkey (2019)	[52]
To examines the sustainable benchmarking of a public transport system.	Practical implementation of public transportation	Fuzzy logic and AHP	✓	India (2019)	[53]
To prioritizing failures for corrective actions in the BRT system.	The maintenance of public transportation	AHP and TOPSIS	-	Turkey (2019)	[54]
To prioritize public transportation projects in Kırıkkale, to enhance urban livability and facilitate transparent decision making for developing cities.	Public transportation planning	AHP and TOPSIS	-	Turkey (2020)	[41]
To develop and implement effective mobility measures for sustainable transport systems.	Promoting sustainable mobility	COPRAS, TOPSIS, ARAS, EDAS, and WAM	-	Lithuania (2020)	[42]
To select electric buses for sustainable transportation.	Selection of sustainable transportation	AHP and TOPSIS	-	Turkey (2020)	[43]
To determine the criteria of the public transport infrastructure that have the most influence on passenger satisfaction.	Urban public transportation planning	Delphi and TOPSIS	-	Lithuania (2020)	[55]
To determine the significant supply quality criteria of public transportation.	Public transportation planning	AHP and BWM	-	Jordan (2020)	[56]
To provide a comprehensive method of evaluation for public transportation in Ciudad Juárez, Chihuahua.	Public transportation evaluation	CODAS and Pythagorean fuzzy sets	✓	Mexico (2021)	[57]
To evaluate sustainable public transportation in Tehran.	Public transportation evaluation	FBWM and MABAC	✓	Iran (2021)	[58]
To assess the quality of public transport services in Budapest, Hungary.	Public transport decision making	FAHP and FTOPSIS	✓	Hungary (2021)	[59]
To assess public transport modes in Kampala, Uganda, from a transportation operator’s perspective.	Public transportation planning	ANP, and ÉLECTRE III	-	Uganda (2022)	[60]
To investigate performance measures and propose recommendations for the sustainable public transport sector in India.	Evaluating the quality of public transport	TOPSIS, VIKOR, ÉLECTRE, Delphi, and FAHP	✓	India (2022)	[35]
To enhance public bus transport, prioritize improvements for citizen welfare and government investments.	Enhancement of public transportation	AHP and ANP	-	Hungary (2023)	[61]
To assess and enhance public transportation in Budapest, Hungary.	Public transportation evaluation	BWM, AHP, and MOORA	-	Hungary (2023)	[62]
To prioritize policy measure options for SULT planning and development in small-town areas of developing countries.	Prioritization of SULT policy measures	AHP and TOPSIS	-	Thailand (2023)	[44]
To determine the most suitable LRT for urban transportation systems and create efficient vehicle fleets.	Selection of public transportation	BWM and WASPAS’PH	-	Turkey (2024)	[63]

Remark: * Full descriptions of all abbreviations used in this study are presented in

Table A1. List of abbreviations.

Abbreviations	Full Names
AHP	Analytic Hierarchy Process
ANP	Analytic Network Process
ARAS	Additive Ratio Assessment
BI	Being Implemented
BNP	Best Non-fuzzy Priority
BRT	Bus Rapid Transit
BWM	Best Worst Method
CI	Consistency Index
COI	Completely Implemented
CODAS	A New Combinative Distance-Based Assessment
COPRAS	Complex Proportional Assessment
CR	Consistency Ratio
DSF	Decision Support Framework
EDAS ÉLECTRE	Evaluation Based on Distance from Average Solution ÉLimination Et Choix Traduisant la REalité
EWM	Entropy Weight Method
FAHP	Fuzzy Analytic Hierarchy Process
FBWM	Fuzzy Best Worst Method
FSM FTOPSIS	Fuzzy Scoring Method Fuzzy Technique for Order of Preference by Similarity to Ideal Solution
GCR	Group Consistency Ratio
HMADM	Hybrid Multi-Attribute Decision Making
KB	Knowledge-Based
KKB	Khon Kaen City Bus
KKC	Khon Kaen City
KonSULT	Knowledgebase on Sustainable Urban Land Use and Transport
LRT	Light Rail Transit
MABAC	Multi-Attributive Border Approximation Area Comparison
MADM	Multi-Attribute Decision Making
MOORA	Multi-Objective Optimization on the Basis of Ratio Analysis
NIS	Negative Ideal Solution
OP	On-Planning
OTP	Office of Transport and Traffic Policy and Planning
PIS	Positive Ideal Solution
RCI SD	Random Consistency Index Sustainable Development
SULT	Sustainable Urban Land Use and Transport
SUMPs	Sustainable Urban Mobility Plans
SUPT	Sustainable Urban Public Transport
SUPTM	Sustainable Urban Public Transport Manual
TFNs	Triangular Fuzzy Numbers
TOD	Transit-Oriented Development
TOPSIS VIKOR	Technique for Order of Preference by Similarity to Ideal Solution VIseKriterijumska Optimizacija I Kompromisno Resenje
WAM	Weighted Average Method
WASPAS’PH	Power–Heronian Weighted Aggregated Sum Product Assessment

in Appendix A.

Table 1. HMADM applications for ranking and evaluating the SULT and SUPT sectors.

Objectives	Applications	Applied *	Fuzzy Environment	Area (Year)	References
To introduce KonSULT, a decision support tool, which identifies synergistic combinations of policy instruments to generate effective policy packages.	Generating and ranking of SULT policy measure	Weighting and scoring methods	-	Europe (2012)	[17]
To introduce a decision support procedure for analyzing and generating consensus among various stakeholders in a transit development.	Utilization of the decision support procedure	FAHP and interval AHP	✓	Turkey (2019)	[52]
To examines the sustainable benchmarking of a public transport system.	Practical implementation of public transportation	Fuzzy logic and AHP	✓	India (2019)	[53]
To prioritizing failures for corrective actions in the BRT system.	The maintenance of public transportation	AHP and TOPSIS	-	Turkey (2019)	[54]
To prioritize public transportation projects in Kırıkkale, to enhance urban livability and facilitate transparent decision making for developing cities.	Public transportation planning	AHP and TOPSIS	-	Turkey (2020)	[41]
To develop and implement effective mobility measures for sustainable transport systems.	Promoting sustainable mobility	COPRAS, TOPSIS, ARAS, EDAS, and WAM	-	Lithuania (2020)	[42]
To select electric buses for sustainable transportation.	Selection of sustainable transportation	AHP and TOPSIS	-	Turkey (2020)	[43]
To determine the criteria of the public transport infrastructure that have the most influence on passenger satisfaction.	Urban public transportation planning	Delphi and TOPSIS	-	Lithuania (2020)	[55]
To determine the significant supply quality criteria of public transportation.	Public transportation planning	AHP and BWM	-	Jordan (2020)	[56]
To provide a comprehensive method of evaluation for public transportation in Ciudad Juárez, Chihuahua.	Public transportation evaluation	CODAS and Pythagorean fuzzy sets	✓	Mexico (2021)	[57]
To evaluate sustainable public transportation in Tehran.	Public transportation evaluation	FBWM and MABAC	✓	Iran (2021)	[58]
To assess the quality of public transport services in Budapest, Hungary.	Public transport decision making	FAHP and FTOPSIS	✓	Hungary (2021)	[59]
To assess public transport modes in Kampala, Uganda, from a transportation operator’s perspective.	Public transportation planning	ANP, and ÉLECTRE III	-	Uganda (2022)	[60]
To investigate performance measures and propose recommendations for the sustainable public transport sector in India.	Evaluating the quality of public transport	TOPSIS, VIKOR, ÉLECTRE, Delphi, and FAHP	✓	India (2022)	[35]
To enhance public bus transport, prioritize improvements for citizen welfare and government investments.	Enhancement of public transportation	AHP and ANP	-	Hungary (2023)	[61]
To assess and enhance public transportation in Budapest, Hungary.	Public transportation evaluation	BWM, AHP, and MOORA	-	Hungary (2023)	[62]
To prioritize policy measure options for SULT planning and development in small-town areas of developing countries.	Prioritization of SULT policy measures	AHP and TOPSIS	-	Thailand (2023)	[44]
To determine the most suitable LRT for urban transportation systems and create efficient vehicle fleets.	Selection of public transportation	BWM and WASPAS’PH	-	Turkey (2024)	[63]

Remark: * Full descriptions of all abbreviations used in this study are presented in Table A1 in Appendix A.

3. Materials and Case Study

3.1. Research Methodology

A flowchart illustrating the research methodology is presented in Figure 1. Each step is briefly described as follows: (1) definition of key objectives: the aim of this research was previously defined in Section 1; (2) literature reviews: various research articles and studies in SUMP, SUPT, SUPTM, KonSULT software, several MADM, and HMADM applications were reviewed; (3) select study area: a medium-sized city (KKC) in a developing country (Thailand) was selected; (4) site survey and data collection: basic information regarding the proposed SUPT policy measures and their current implementation stages for KKC was gathered; (5) selection of experts: 26 human experts in various relevant fields were selected (each expert was interviewed directly to obtain their expertise on the relative weights of all determined criteria); (6) development of the DSF: a combination of the SUPTM policy options, the KonSULT KB, and the HMADM method (including the FAHP, FSM, and TOPSIS); (7) prioritizing all potential policy options: prioritization of all options was achieved using the HMADM; and finally (8) the recommendation of suitable actions and implementations: the ranking order of all proposed policy options and their current implementation stages will be considered to provide suggestions for appropriate actions and implementations.

3.2. Selection of the Study Area

Khon Kaen City (KKC), with 384,000 residents, is a medium-sized city in the Khon Kaen province located in the central part of the northeastern region of Thailand. It is located approximately 445 km from Bangkok. KKC is a dynamic developing city covering approximately 228 km², as shown in Figure 2. KKC is becoming a regional hub for transport and logistics, medical, educational, financial, administrative, conventions and exhibitions, smart cities, and low-carbon cities [64]. Khon Kaen is witnessing several significant transport infrastructure projects aimed at fostering its economic growth and many strategic hubs. For instance, such projects are as follows: the east–west economic corridor [65]; the high-speed rail project; the double-track rail construction; and the motorway (M6) project, [66,67,68]. KKC was selected as the study area because it is one of most vital strategic regional cities of Thailand. The general historical trends of the numbers of population, employment, and registered vehicles gradually increased from 2013 to 2022, as shown in Figure 3 [20,32].

Figure 2. Study area.

Figure 2. Study area.

Figure 3. Historical trends of the numbers of population, employment, and registered vehicles, and the GPP in KKC [20,32].

Figure 3. Historical trends of the numbers of population, employment, and registered vehicles, and the GPP in KKC [20,32].

Currently, paratransit is the predominant mode of the urban public transport system in KKC. It consists of several types of paratransit systems, including motorcycles taxis, tricycle taxis, tuk-tuks, and songthaew. The limited Khon Kaen city bus (KKB) and van routes have also provided their services to the public. KKC has been suffering from the drastic and uncontrollable expansion of built-up areas away from the central business district since 1990 [69]. This primary land use crisis, along with the current lack of efficient and sufficient public transportation infrastructure and services, the relatively rapid growth of economic development, urbanization, population, employment, and the travel demand, has adversely leaded to traffic jams, road accidents, environmental impacts, excessive energy use, and climate change.

3.3. Site Survey and Data Collection

Recently, OTP [20] developed a sustainable urban public transport manual (SUPTM) for 78 regional cities across Thailand. The SUPTM manual can recommend a set of various policy measures for SUPT planning. However, SUPTM has never offered any guidelines for ranking and choosing suitable policy measures for SUPT planning in such regional cities. According to the SUPTM manual [20] and various studies [32,70,71,72,73], all 34 policy measures available in the SUPTM manual and various policy measures proposed for KKC were illustrated and classified in accordance with KonSULT KB [12], as shown in Table 2. In addition, the proposed policy measures are categorized according to their current implementation stages (including the completely implemented (COI), being implemented (BI), and on-planning (OP) stages), as shown in Table 2. For ease of analysis and comprehension, the least implementation stages of each proposed policy option were used as a representative of the implementation stages of such an option. Figure 4 displays examples of the measures implemented in the study area.

Table 2. The SUPT policy options and their implementation stages in KKC.

Options Categories	SUPTM Policy Options [20]	SUPT Policy Options [32,70,71,72,73,74,75]	Implementation Stages *			Final Stages ***
			COI	BI	OP
Attitudinal and behavior	Promotional activities (O1)	Education, campaigning, and public relations for KKB, BRT, LRT, and Tram promotion	⨯	-	-	COI
	Telecommunications (O2)	Internet of Things (IoT) for KKB, BRT, LRT, and Tram	-	⨯	-	BI
Infrastructure	Bus rapid transit (BRT) (O3)	BRT development	-	-	⨯	OP
	New rail stations and lines (O4)	New railway line and stations	⨯	-	-	COI
	Park and ride (O5)	Park and ride for BRT, LRT, and Tram	-	-	⨯	OP
	Pedestrian areas and routes (O6)	Provision of pedestrian routes and networks	-	⨯	-	BI
	Terminals and interchanges (O7)	New bus terminal construction	⨯	-	-	COI
	Trams and light rail transit (LRT) (O8)	LRT line development	-	-	⨯	OP
Information provision	Barrier-free Mobility (O9)	Use of the barrier-free mobility vehicles for KKB	⨯	-	-	OP
		Use of the barrier-free mobility vehicles for BRT, LRT, and Tram	-	-	⨯
	Conventional signs and markings (O10)	Installation of typical traffic signs and markings	⨯	-	-	COI
	Conventional timetable and service information (O11)	Timetables and services information for KKB	⨯	-	-	OP
		Timetable and service information for BRT, LRT, and Tram	-	-	⨯
	Crowd sourcing (O12)	Mobile application for LRT	-	-	⨯	OP
		Mobile application for KKB	⨯	-	-
	In-vehicle guidance systems (O13)	In-vehicle navigation and tracking system for KKB	⨯	-	-	OP
		In-vehicle navigation and tracking system for BRT, LRT, and Tram	-	-	⨯
	Real-time passenger information (O14)	Real time traveler information systems for KKB	⨯	-	-	OP
		Real time traveler information systems for BRT, LRT, and Tram	-	-	⨯
Land use	Development density and mix (O15)	TOD development along the LRT lines	-	-	⨯	OP
	Land use to support public transport (O16)	Land use control to support LRT development	-	-	⨯	OP
Management and service	Bus fleet management systems (O17) **	Bus routes and services management	-	⨯	-	BI
	Bus priorities (O18) **	Installation of KKB, BRT, LRT, and Tram priority system	-	-	⨯	OP
	Bus regulation (O19) **	Establishment of the Khon Kaen transit system (KKTS) organization to regulate KKB, BRT, LRT, and Tram	⨯	-	-	COI
	Bus services (O20) **	KKB service quality improvement	⨯	-	-	OP
		Provision of BRT, LRT, and Tram services	-	-	⨯
	Conventional traffic management (O21)	Traditional traffic management	⨯	-	-	COI
	Cycle parking and storage (O22)	Provision of cyclist facilities	-	⨯	-	BI
	Intelligent transport systems (ITSs) (O23)	ITS signals installations	-	⨯	-	BI
	New rail services (O24)	New tram service development	-	-	⨯	OP
	Parking controls (O25)	Parking controls	⨯	-	-	COI
	Pedestrian crossing facilities (O26)	Provision of pedestrians crossing facilities	-	⨯	-	BI
	Physical restrictions (O27)	Road space reduction	-	-	⨯	OP
	Urban traffic control (O28)	Urban traffic control	-	⨯	-	BI
Pricing	Fare levels (O29)	Fare levels for BRT, LRT, and Tram	-	-	⨯	OP
		Fare level for KKB	⨯	-	-
	Fare structures (O30)	Fare structures for BRT, LRT, and Tram	-	-	⨯	OP
		Fare structure for KKB	⨯	-	-
	Fuel taxes (O31)	-	-	-	-	-
	Integrated ticketing (O32)	Integrated ticketing for LRT	-	-	⨯	OP
	Parking charges (O33)	-	-	-	-	-
	Road user charging (O34)	-	-	-	-	-
Total	34					31

Remark: * The current public transport implementation stages in the study area; ** Bus = all public transport modes. *** COI = Completely implemented stage; BI = Being implemented stage; OP = On-planning stage; - = Not proposed in the SUPT planning.

Table 2. The SUPT policy options and their implementation stages in KKC.

Options Categories	SUPTM Policy Options [20]	SUPT Policy Options [32,70,71,72,73,74,75]	Implementation Stages *			Final Stages ***
			COI	BI	OP
Attitudinal and behavior	Promotional activities (O1)	Education, campaigning, and public relations for KKB, BRT, LRT, and Tram promotion	⨯	-	-	COI
	Telecommunications (O2)	Internet of Things (IoT) for KKB, BRT, LRT, and Tram	-	⨯	-	BI
Infrastructure	Bus rapid transit (BRT) (O3)	BRT development	-	-	⨯	OP
	New rail stations and lines (O4)	New railway line and stations	⨯	-	-	COI
	Park and ride (O5)	Park and ride for BRT, LRT, and Tram	-	-	⨯	OP
	Pedestrian areas and routes (O6)	Provision of pedestrian routes and networks	-	⨯	-	BI
	Terminals and interchanges (O7)	New bus terminal construction	⨯	-	-	COI
	Trams and light rail transit (LRT) (O8)	LRT line development	-	-	⨯	OP
Information provision	Barrier-free Mobility (O9)	Use of the barrier-free mobility vehicles for KKB	⨯	-	-	OP
		Use of the barrier-free mobility vehicles for BRT, LRT, and Tram	-	-	⨯
	Conventional signs and markings (O10)	Installation of typical traffic signs and markings	⨯	-	-	COI
	Conventional timetable and service information (O11)	Timetables and services information for KKB	⨯	-	-	OP
		Timetable and service information for BRT, LRT, and Tram	-	-	⨯
	Crowd sourcing (O12)	Mobile application for LRT	-	-	⨯	OP
		Mobile application for KKB	⨯	-	-
	In-vehicle guidance systems (O13)	In-vehicle navigation and tracking system for KKB	⨯	-	-	OP
		In-vehicle navigation and tracking system for BRT, LRT, and Tram	-	-	⨯
	Real-time passenger information (O14)	Real time traveler information systems for KKB	⨯	-	-	OP
		Real time traveler information systems for BRT, LRT, and Tram	-	-	⨯
Land use	Development density and mix (O15)	TOD development along the LRT lines	-	-	⨯	OP
	Land use to support public transport (O16)	Land use control to support LRT development	-	-	⨯	OP
Management and service	Bus fleet management systems (O17) **	Bus routes and services management	-	⨯	-	BI
	Bus priorities (O18) **	Installation of KKB, BRT, LRT, and Tram priority system	-	-	⨯	OP
	Bus regulation (O19) **	Establishment of the Khon Kaen transit system (KKTS) organization to regulate KKB, BRT, LRT, and Tram	⨯	-	-	COI
	Bus services (O20) **	KKB service quality improvement	⨯	-	-	OP
		Provision of BRT, LRT, and Tram services	-	-	⨯
	Conventional traffic management (O21)	Traditional traffic management	⨯	-	-	COI
	Cycle parking and storage (O22)	Provision of cyclist facilities	-	⨯	-	BI
	Intelligent transport systems (ITSs) (O23)	ITS signals installations	-	⨯	-	BI
	New rail services (O24)	New tram service development	-	-	⨯	OP
	Parking controls (O25)	Parking controls	⨯	-	-	COI
	Pedestrian crossing facilities (O26)	Provision of pedestrians crossing facilities	-	⨯	-	BI
	Physical restrictions (O27)	Road space reduction	-	-	⨯	OP
	Urban traffic control (O28)	Urban traffic control	-	⨯	-	BI
Pricing	Fare levels (O29)	Fare levels for BRT, LRT, and Tram	-	-	⨯	OP
		Fare level for KKB	⨯	-	-
	Fare structures (O30)	Fare structures for BRT, LRT, and Tram	-	-	⨯	OP
		Fare structure for KKB	⨯	-	-
	Fuel taxes (O31)	-	-	-	-	-
	Integrated ticketing (O32)	Integrated ticketing for LRT	-	-	⨯	OP
	Parking charges (O33)	-	-	-	-	-
	Road user charging (O34)	-	-	-	-	-
Total	34					31

Remark: * The current public transport implementation stages in the study area; ** Bus = all public transport modes. *** COI = Completely implemented stage; BI = Being implemented stage; OP = On-planning stage; - = Not proposed in the SUPT planning.

Figure 4. Some examples of important SUPT projects in KKC [32,68,70,71,73,74,75].

Figure 4. Some examples of important SUPT projects in KKC [32,68,70,71,73,74,75].

3.4. Selection of Experts

In this study, 26 experts were selected and interviewed directly to determine the relative weights of the decision criteria affecting the prioritization of SUPT policy measure options in KKC. These experts were selected based on the following criteria: Each expert must directly be involved in at least two SULT or SUPT projects in KKC. Each expert must have at least five years of professional or practical experience and expertise in urban land use and transport, urban public transport planning and development, town planning, and urban design. Nine experts (34%) were academics in urban land use and transport planning and development, six experts (23%) were policy decision makers in urban public transport planning who worked in government organizations, eight experts (31%) were academic researchers in urban design and town planning, and three experts (12%) were policy decision makers in urban design and town planning in government agencies.

3.5. Development of an Innovative Decision Support Framework

An innovative DSF consisting of 3 main components included (1) the SUPT option generation; (2) the options’ performance score evaluation; and (3) the HMADM method (FAHP, FSM, and TOPSIS).

3.5.1. The SUPT Policy Option Generation

Based on the SUPTM manual [20], 34 policy measure options were adopted for SUPT planning in KKC, Thailand. According to several SUPT studies in KKC [32,70,71,72,73], several SUPT policy options were proposed for the city center area of KKC. A few of these proposed policy options were completely implemented (COI); some options are being implemented (BI), and several options are in the on-planning (OP) stage.

3.5.2. The Options’ Performance Score Evaluation

KonSULT KB [12] was adopted to assign the performance (contribution) scores of each policy option in association with all 7 objectives. Such performance scores ranging from −5 (strongest possible negative contribution) to +5 (the strongest possible positive contribution) were achieved from the practical and professional knowledge and expertise of many human experts from various European countries.

3.5.3. The HMADM Technique

• The fundamental basis of the fuzzy set theory

The notion of the fuzzy set theory was first introduced by Zadeh [76], wherein a fuzzy set was defined as a collection of components along with their respective degrees of belonging, referred to as grades of membership. The fuzzy set theory is commonly utilized to tackle uncertain, imperfect, and vague (fuzzy) information, and is considered a generalized set theory allowing the theoretical flexibility of the conventional set theory.

2.

The concept of fuzzy numbers

Fuzzy numbers can be used to identify the range of values for a specific criterion or performance score. A single linguistic score can then be transformed into a fuzzy number [77]. Triangular fuzzy numbers (TFNs) are commonly used because of their computational simplicity and practicality in the fuzzy environment [78]. TFNs can be defined using three numbers (l, m, and u), where l, m, and u represent the minimum, most probable, and maximum possible values, respectively [77]. The mathematical formulation of fuzzy number x with membership function µ_A(x) is provided in Figure 5 [77].

Figure 5. Triangular fuzzy number (TFN) [77].

Figure 5. Triangular fuzzy number (TFN) [77].

3.

Fuzzy Analytical Hierarchy Process (FAHP)

Laarhoven and Pedrycz [26] proposed a FAHP method by applying TFNs to a pairwise comparison matrix [48]. The procedural steps of FAHP applications are summarized as follows:

Step 1: Creating the hierarchical structure and identifying all criteria.

Step 2: In this step, each expert might generate a pairwise comparison matrix of all decision criteria in a defined hierarchy structure using linguistic (fuzzy) scales. The nine-point ratio scale system [79,80] was applied in the pairwise comparison process of the FAHP to transform linguistic (fuzzy) scales into their corresponding TFNs. Based on Saaty’s ratio scale system [45], the typical nine-point ratio scales can be defined as follows: 1 (equal importance), 3 (moderate), 5 (strong), 7 (very strong), and 9 (extremely strong). The remaining four intermediate scales (2, 4, 6, and 8) between those scale intervals are considered intermediate scales. These Saaty’s ratio scales and their corresponding TFNs, as well as the reciprocal TFNs, are shown in

Table 3. Linguistic ratio scales and their corresponding TFNs for pairwise comparisons of all determined criteria [79,80].

Saaty’s Ratio Scale	Definitions	TFNs	Reciprocal TFNs
1	Equal	(1, 1, 1)	(1, 1, 1)
3	Moderate	(2, 3, 4)	(1/4, 1/3, 1/2)
5	Strong	(4, 5, 6)	(1/6, 1/5, 1/4)
7	Very strong	(6, 7, 8)	(1/8, 1/7, 1/6)
9	Extremely strong	(9, 9, 9)	(1/9, 1/9, 1/9)
2	Intermediate scales	(1, 2, 3)	(1/3, 1/2, 1)
4		(3, 4, 5)	(1/5, 1/4, 1/3)
6		(5, 6, 7)	(1/7, 1/6, 1/5)
8		(7, 8, 9)	(1/9, 1/8, 1/7)

.

Step 3: Determining the fuzzy relative weight or fuzzy synthetic extent of each criterion, as shown in Equations (1)–(4), by adopting Chang’s method [81]. Let X = {x₁, x₂, x₃, …, x_n} be a set goal. The analysis for each goal, x_i, can be performed separately for each criterion [77]. Accordingly, the m extent value for each criterion can be achieved as follows: M¹_xi, M²_xi, M³_xi, …, M^m_xi, where x_i (i = 1, 2, 3, …, n) is the goal set, and M^j_xi (j = 1, 2, 3, …, m) are all TFNs. The value of the fuzzy synthetic extent (S_i) corresponding to the ith criterion can be expressed as Equation (1).

$$S_i={{\displaystyle {\sum }_{j=1}^m{M}_{x_i}^j\otimes \left[{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^m{M}_{x_i}^j}}\right]}}^{-1}$$

(1)

To estimate ${\displaystyle {\sum }_{j=1}^m{M}_{x_i}^j}$, a fuzzy addition operation of the m extent is applied to a matrix, as shown in Equation (2).

$${\displaystyle {\sum }_{j=1}^m{M}_{x_i}^j}=\left({\displaystyle {\sum }_{j=1}^m{l}_j,}{\displaystyle {\sum }_{j=1}^m{m}_j,}{\displaystyle {\sum }_{j=1}^m{u}_j}\right)$$

(2)

where l, m, and u are the minimum, most probable, and maximum values, respectively. As shown in Equation (3), ${\left[{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^m{M}_{x_i}^j}}\right]}^{-1}$ can be estimated by computing the inverse of the vector.

$${\left[{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^m{M}_{x_i}^j}}\right]}^{-1}=\left(1/{\displaystyle {\sum }_{i=1}^n{u}_i,1/{\displaystyle {\sum }_{i=1}^n{m}_i,}1/{\displaystyle {\sum }_{i=1}^n{l}_i}}\right)$$

(3)

Step 4: Estimating the relative weight of each criterion involves defuzzifying the fuzzy relative weights of the determined criteria. To defuzzify these relative weights, the defuzzification method presented in Equation (4) is adopted to achieve the best non-fuzzy priority (BNP) or crisp relative weights of the criteria, where i = 1, 2, and 3 [77].

$$BN{P}_{S_i}=\frac{[(u_{s_i})-(l_{s_i})+(m_{s_i})-(l_{s_i})]}{3}+l_s$$

(4)

Using these BNP values, all criteria can be prioritized by considering the magnitude of their relative weights. The most important criterion was the largest BNP value, and the least important was the lowest BNP value. The relative weights (w_i) of each criterion can be standardized, as shown in Equation (5).

$$w_i=BN{P}_{s_i}/{\displaystyle {\sum }_{i=1}^{n}BN{P}_{s_i}}$$

(5)

The local relative weights of all decision criteria contained in each hierarchical level of the entire hierarchical structure were estimated. The FAHP typically applies the “principle of hierarchical composition” to aggregate the local relative weights of all criteria in all hierarchical levels to achieve the global relative weights of each decision criterion at the lowest hierarchical level [45].

The consistency Index (CI) was then estimated by using an equation (CI = (λ_max − n)/(n − 1)) to determine the consistency of the square matrix A and the derived λ_max is the largest right eigenvalue of the pairwise comparison matrix. The consistency ratio (CR) was then computed as the ratio between the CI and the random consistency index (RCI), as shown in Table 4 [46]. A CR less than or equal to 0.10 is commonly acceptable. The geometric mean method (GMM) was applied to combine all expert judgments to achieve the group judgement [82].

Table 4. Random consistency index (RCI) [83].

n	3	4	5	6	7	8	9
RCI	0.52	0.89	1.11	1.25	1.35	1.40	1.45

Table 4. Random consistency index (RCI) [83].

n	3	4	5	6	7	8	9
RCI	0.52	0.89	1.11	1.25	1.35	1.40	1.45

4.

The Fuzzy Scoring Method (FSM)

Chen and Hwang [27] initially established a simplification method called the fuzzy scoring method (FSM) to transform linguistic (fuzzy) scores into numerical (crisp) scores. In the FSM, based on the calculations of the left and right utility scores of each defined TFN [27], the total utility score $({\mu }_T(i))$ of a TFN can then be estimated. In this study, all performance (fuzzy) scores of each policy measure option under each criterion (objective) are defined by adopting their corresponding TFNs. The greater the total utility score, the better the TFNs. Based on Liu and Chen [84], when the TFN (M_TFN = (l, m, u)) is defined, the total utility score of such a TFN $(M_T(i))$ can be simply calculated from Equation (6).

$$M_T(i)=\left([u/(1-m+u)]+[m/(m-l+1)]\right)\times 0.5$$

(6)

5.

TOPSIS Method

TOPSIS was initially developed by Hwang and Yoon [29] to determine the best option based on the principle of compromise. In TOPSIS, the best option possesses the minimum Euclidean distance (D) from the positive ideal solution (PIS) and the maximum D from the negative ideal solution (NIS) [85]. The computational steps of TOPSIS are as follows:

Step 1: A normalized decision matrix is calculated, where the normalized value (n_ij) is estimated using Equation (7); PS_ij is the performance score of option under criterion, i is an option, n is the total number of options, j is the criterion, and m is the total number of criteria.

$$n_{ij}=P{S}_{ij}/\sqrt{{\displaystyle {\sum }_i^{n}P{S}_{ij}^2}};i=1, 2, \dots , n; j=1, 2, \dots , m.$$

(7)

Step 2: A weighted normalized decision matrix is computed, where the weighted normalized value, wn_ij, is estimated from w_j × n_ij, where w_j is the relative weight of criterion.

Step 3: Consider PIS (or P_j⁺) and NIS (or N_j⁻), which are determined by Equations (8) and (9), respectively, where J1 is associated with beneficial criteria (i.e., the higher the value, the better the performance), and J2 is associated with non-beneficial criteria (i.e., the lower the value, the better the performance).

$$P_j^{+}=\{(\underset{i}{\max }w{n}_{ij}\left|j\in J_1),(\underset{i}{\min }w{n}_{ij}\right|j\in J_2)\};i=1, 2, \dots , n; j=1, 2, \dots , m.$$

(8)

$$N_j^{-}=\{(\underset{i}{\min }w{n}_{ij}\left|j\in J_1),(\underset{i}{\max }w{n}_{ij}\right|j\in J_2)\};i=1, 2, \dots , n; j=1, 2, \dots , m.$$

(9)

Step 4: The separation from either PIS or NIS for each option is considered. Separation values can be defined as D, and the separation of each option from the PIS (D_i⁺) is estimated using Equation (10). Similarly, the separation of each option from the NIS (D_i⁻) is quantified by Equation (11).

$$D_i^{+}=\sqrt{{\displaystyle {\sum }_{j=1}^m{(w{n}_{ij}-P_j^{+})}^2}};i=1, 2, \dots , n.$$

(10)

$$D_i^{-}=\sqrt{{\displaystyle {\sum }_{j=1}^m{(w{n}_{ij}-N_j^{-})}^2}};i=1, 2, \dots , n.$$

(11)

Step 5: The relative closeness (RC_i) to the ideal solution or composite score (CS_i) is calculated by RC_i = CS_i = D_i⁻/(D_i⁺ + D_i⁻) when i = 1, 2, 3, …, n.

Step 6: All options are prioritized based on their RC_i or CS_i values, in descending order from the top-priority option to the lowest-priority option. The best option has the minimum D from the PIS and the maximum D from the NIS [42]. This means that the higher the RC_i or CS_i, the superior the option.

4. Results

4.1. Application of the FAHP to Determine the Relative Weights of All Criteria

4.1.1. Organization of Decision Elements into a Hierarchical Structure

Based on the SUPTM manual [20] and the KonSULT KB [12], all decision criteria were identified, and then formulated as a hierarchical structure (as shown in Figure 6) essentially for prioritizing policy measure options in SUPT planning in KKC. The hierarchical structure illustrated the relationships among all the decision criteria in the hierarchical structure, ensuring that the top decision criterion is connected to the lowest decision criteria at the bottom [45]. The principal goal of this decision-making process is shown in level 1. Furthermore, a box immediately underneath level 1 represents all 26 selected experts who were directly interviewed to provide their judgments on the relative weights of both the main and minor decision criteria. Based on the sustainable development (SD) concept, the three principal components of SD, namely (1) economic development, (2) social responsibility, and (3) environmental protection, were defined as the main criteria in level 2. Based on the SUPTM [20] and KonSULT [12], all seven key objectives (including (1) economic efficiency, (2) economic growth, (3) finance, (4) livable streets and neighborhood, (5) equity and social inclusion, (6) safety, and (7) environmental protection) were also identified as seven minor criteria in level 3, corresponding to the three main criteria in level 2. In level 4, all 34 potential SUPT policy measure options were obtained from the SUPTM manual [20]. In addition, between levels 3 and 4, the performance scores (PS_i) of each option (i) contributing to all seven minor criteria (objectives) were achieved from the KonSULT KB [12].

4.1.2. Estimations of Relative Weights of Both Main and Minor Criteria

The FAHP was adopted as a powerful means to acquire practical and professional knowledge and expertise on the relative importance of both the main and minor criteria from the 26 experts. Each expert was directly interviewed to generate several pairwise comparison matrices of both main and minor criteria contained in hierarchical levels 2 and 3, respectively. The nine-point scaling system (as presented in Table 3) adopted in the FAHP was applied to convert the linguistic (fuzzy) scales into their corresponding TFNs. The GMM method introduced by Buckley [82] was subsequently applied to combine each individual judgments of all 26 experts to achieve fuzzy geometric mean pairwise comparison matrices for both the main and minor criteria. Once the TFN pairwise comparison matrices were completely generated by each expert, the fuzzy geometric mean pairwise comparison metrics of each expert or a group of experts, as illustrated in some examples (

Table 5. A group pairwise comparison matrix for the three main criteria.

Main Criteria	(C1)	(C2)	(C3)
Economic Development (C1)	(1, 1, 1)	(0.802, 1.052, 1.344)	(1.016, 1.415, 1.934)
Social Responsibility (C2)	(0.744, 0.951, 1.248)	(1, 1, 1)	(1.140, 1.507, 1.878)
Environmental Protection (C3)	(0.517, 0.707, 0.985)	(0.532, 0.664, 0.877)	(1, 1, 1)

Note: λ_max = 3.0004, GCI = 0.0001, and GCR = 0.0003

,

Table 6. A group pairwise comparison matrix for the three minor criteria under the “Economic development” criterion.

Minor Criteria	(C11)	(C12)	(C13)
Economic Efficiency (C11)	(1, 1, 1)	(1.594, 2.038, 2.566)	(1.782, 2.399, 3.020)
Economic Growth (C12)	(0.390, 0.491, 0.627)	(1, 1, 1)	(0.972, 1.370, 1.843)
Finance (C13)	(0.331, 0.417, 0.561)	(0.543, 0.730, 1.029)	(1, 1, 1)

Note: λ_max = 3.0026, GCI = 0.0009, and GCR = 0.0016.

and

Table 7. A group pairwise comparison matrix for the three minor criteria under the “Social responsibility” criterion.

Minor Criteria	(C21)	(C22)	(C23)
Livable Streets and Neighborhood (C21)	(1, 1, 1)	(0.621, 0.789, 0.998)	(0.303, 0.378, 0.498)
Equity and Social Inclusion (C22)	(1.002, 1.267, 1.611)	(1, 1, 1)	(0.354, 0.451, 0.596)
Safety (C23)	(2.009, 2.649, 3.299)	(1.678, 2.219, 2.827)	(1, 1, 1)

Note: λ_max = 3.0004, GCI = 0.0001, and GCR = 0.0003.

), could be achieved. These matrices could be used to calculate the relative weights of the three main criteria as well as the seven minor criteria. It should be noted that the Human Research and Ethics Committee of Khon Kaen University reviewed and approved this study (HE653281).

4.1.3. Determination of the Global Relative Weights of All Criteria

It was essential to calculate the fuzzy relative weights or fuzzy synthetic extents (S_i) for all environmental criteria when determining their relative weights. The defuzzification method outlined in Equation (1) was used to address the fuzziness of the weights. The goal of this task was to determine the BNP or numerical (crisp) relative weights of both the three main criteria and their corresponding seven minor criteria are presented in

Table 8. Group relative weights of both the main and minor criteria of all 26 experts.

Main and Minor Criteria	Si Lower	Si Medium	Si Upper	Main Criteria			Minor Criteria
				BNP	Final Weight	Rank	BNP	Final Weight	Rank
Economic Development (C1)	0.250	0.373	0.552	0.393	0.375	1
Economic efficiency (C11)	0.345	0.520	0.765				0.543	0.194	3
Economic growth (C12)	0.188	0.276	0.405				0.290	0.103	4
Finance (C13)	0.148	0.205	0.299				0.217	0.078	6
Social Responsibility (C2)	0.256	0.372	0.532	0.385	0.367	2
Livable streets and neighborhoods (C21)	0.148	0.205	0.299				0.210	0.074	7
Equity and social inclusion (C22)	0.149	0.201	0.279				0.263	0.092	5
Safety (C23)	0.181	0.251	0.356				0.572	0.201	2
Environmental Protection (C3)	0.182	0.255	0.369	0.271	0.258	3
Environmental protection (C31)	0.182	0.255	0.369				-	0.258	1
Sum				-	1.000		-	1.000	-

Remark: C31 = C3.

and Figure 7.

As shown in Figure 8, all estimated CR values of each expert for the three main criteria, the three minor criteria under the “economic development” criterion, and the other three minor criteria under the “social responsibility” criterion were less than 0.1. As illustrated in Table 5, Table 6 and Table 7, the estimated GCR values of a group of all 26 experts for the three main criteria, the three minor criteria under the “economic development” criterion, and the other three minor criteria under the “social responsibility” criterion were 0.0003, 0.0016, and 0.0003, respectively. These GCR values were also less than 0.1. This means that the judgments of each expert and the group of all 26 experts were acceptably consistent [45].

4.2. Performance Scores of the Potential Policy Measure Options for All Minor Criteria

In Equation (6), the TFNs were used to determine the performance (fuzzy) scores of each option under all criteria (objectives). As illustrated in Figure 9, each TFN can be defined by assigning a number ranging from zero (not belonging) to one (fully belonging) according to its grade of membership in that element. Furthermore, TFNs can be characterized using three numbers (l, m, and u), where l, m, and u represent the minimum, most probable, and maximum possible values, respectively [76]. For any TFN, in the first (left) interval from l to m, grades of membership increase linearly from zero to one, respectively; in the second (right) interval from m to u, grades of membership decrease linearly from one to zero, respectively. Any element less than l or greater than m is considered infeasible. Using these TFNs in the decision-making process allows us to interpret each TFN score as a measure of how strongly a specific option contributes to each criterion (objective). For instance, if any option is evaluated as the “medium possible positive contribution (PS₉)” score, its corresponding TFN can be represented by M_TFN = (0.7, 0.8, 0.9). All 11 scales of TFNs representing all performance scores are illustrated in Figure 9, and the numerical values of all total utility scores for all 11 performance scores are presented in

Table 9. Estimated utility scores for the eleven-point scale.

Performance Scores (PSi)	Linguistic (Fuzzy) Scores	Components of MTFN(i)			Total Utility Scores $({\mathit{\mu }}_{\mathit{T}}(\mathit{i}))$	Normalized ${\mathit{\mu }}_{\mathit{T}}(\mathit{i})$
		l	m	u
PS1	Strongest possible negative contribution	0.0	0.0	0.1	0.0455	0.0476
PS2	Strong possible negative contribution	0.0	0.1	0.2	0.1364	0.1429
PS3	Medium possible negative contribution	0.1	0.2	0.3	0.2273	0.2381
PS4	Weak possible negative contribution	0.2	0.3	0.4	0.3182	0.3333
PS5	Weakest possible negative contribution	0.3	0.4	0.5	0.4091	0.4286
PS6	No contribution	0.4	0.5	0.6	0.5000	0.5238
PS7	Weakest possible positive contribution	0.5	0.6	0.7	0.5909	0.6190
PS8	Weak possible positive contribution	0.6	0.7	0.8	0.6818	0.7143
PS9	Medium possible positive contribution	0.7	0.8	0.9	0.7727	0.8095
PS10	Strong possible positive contribution	0.8	0.9	1.0	0.8636	0.9048
PS11	Strongest possible positive contribution	0.9	1.0	1.0	0.9545	1.0000

[86].

All 34 potential SUPT policy options contained in the SUPTM manual [20] (Table 2) were determined and prioritized in this research. The performance (contribution) scores (contained in the KonSULT KB) of each policy measure option, corresponding to all seven minor criteria (objectives), were adopted. The performance scores of all 34 potential policy measure options for each of the seven minor (objectives) criteria are illustrated in

Table 10. Performance scores of all 34 potential policy measure options associated with the seven minor criteria.

Option Codes	Names of Options	Performance (Contribution) Scores of Each Option for All Minor Criteria (Objectives) *
		C11	C12	C13	C21	C22	C23	C31
O1	Promotional activities	0.7143	0.5238	0.4286	0.7143	0.7143	0.6190	0.9048
O2	Telecommunications	0.8095	0.5238	0.4286	0.7143	0.5238	0.5238	0.7143
O3	Bus rapid transit	0.9048	0.6190	0.3333	0.7143	0.7143	0.7143	0.7143
O4	New rail stations and lines	0.8095	0.7143	0.2381	0.7143	0.7143	0.7143	0.7143
O5	Park and ride	0.7143	0.5238	0.3333	0.8095	0.5238	0.6190	0.6190
O6	Pedestrian areas and routes	0.8095	0.8095	0.5238	1.0000	1.0000	0.9048	0.7143
O7	Terminals and interchanges	0.8095	0.7143	0.2381	0.6190	0.7143	0.5238	0.6190
O8	Trams and light rail	0.7143	0.7143	0.1429	0.7143	0.7143	0.7143	0.7143
O9	Barrier-free mobility	0.5238	0.5238	0.5238	0.5238	0.9048	0.8095	0.5238
O10	Conventional signs and markings	0.7143	0.5238	0.5238	0.5238	0.5238	0.7143	0.6190
O11	Conventional timetable and service information	0.7143	0.5238	0.6190	0.6190	0.7143	0.6190	0.6190
O12	Crowd sourcing	0.8095	0.5238	0.7143	0.5238	0.4286	0.6190	0.5238
O13	In-vehicle guidance systems	0.7143	0.5238	0.4286	0.3333	0.6190	0.6190	0.7143
O14	Real-time passenger information	0.7143	0.5238	0.3333	0.6190	0.6190	0.6190	0.6190
O15	Development density and mix	0.8095	0.8095	0.5238	0.9048	0.9048	0.8095	0.7143
O16	Land use to support public transport	0.8095	0.7143	0.7143	0.9048	0.7143	0.8095	0.9048
O17	Bus fleet management systems	0.7143	0.5238	0.6190	0.6190	0.6190	0.6190	0.6190
O18	Bus priorities	0.6190	0.6190	0.5238	0.7143	0.7143	0.7143	0.7143
O19	Bus regulation	0.7143	0.7143	0.6190	0.6190	0.8095	0.7143	0.6190
O20	Bus services	0.6190	0.5238	0.4286	0.6190	0.8095	0.6190	0.7143
O21	Conventional traffic management	0.7143	0.5238	0.4286	0.2381	0.3333	0.8095	0.2381
O22	Cycle parking and storage	0.7143	0.6190	0.3333	0.8095	0.7143	0.5238	0.7143
O23	Intelligent transport systems	0.9048	0.7143	0.3333	0.5238	0.7143	0.9048	0.8095
O24	New rail services	0.7143	0.7143	0.3333	0.6190	0.8095	0.7143	0.7143
O25	Parking controls	0.7143	0.7143	0.3333	0.7143	0.6190	0.7143	0.7143
O26	Pedestrian crossing facilities	0.5238	0.7143	0.3333	0.8095	0.7143	0.8095	0.5238
O27	Physical restrictions	0.3333	0.5238	0.4286	0.6190	0.4286	0.5238	0.5238
O28	Urban traffic control	1.0000	0.5238	0.2381	0.5238	0.7143	0.6190	0.7143
O29	Fare levels	0.7143	0.6190	0.3333	0.7143	0.8095	0.6190	0.7143
O30	Fare structures	0.8095	0.7143	0.9048	0.7143	0.7143	0.7143	0.7143
O31	Fuel taxes	0.7143	0.6190	0.7143	0.6190	0.4286	0.7143	0.7143
O32	Integrated ticketing	0.9048	0.5238	0.3333	0.6190	0.9048	0.7143	0.7143
O33	Parking charges	0.8095	0.4286	0.7143	0.7143	0.7143	0.7143	0.7143
O34	Road user charging	0.9048	0.5238	0.9048	0.8095	0.8095	0.7143	0.8095

Remark: * C11 = Economic efficiency, C12 = Economic growth, C13 = Finance, C21 = Livable streets and neighborhoods, C22 = Equity and social inclusion, C23 = Safety, and C31 = C3 = Environmental protection; The meaning of the score for each color shade in the table is detailed as follows:

0.0476	=PS1	=Strongest possible negative contribution	0.6190	=PS7	=Weakest possible positive contribution
0.1429	=PS2	=Strong possible negative contribution	0.7143	=PS8	=Strongest possible negative contribution
0.2381	=PS3	=Medium possible negative contribution	0.8095	=PS9	=Weak possible positive contribution
0.3333	=PS4	=Weak possible negative contribution	0.9048	=PS10	=Strong possible positive contribution
0.4286	=PS5	=Weakest possible negative contribution	1.0000	=PS11	=Strongest possible positive contribution
0.5238	=PS6	=No contribution

.

4.3. Prioritization of All Options Using TOPSIS

After the defuzzification process, the numerical values of both the relative weights (w_j) of each minor criterion and the performance scores (PS_i) of each option (i) associated with all seven minor criteria (j) were obtained. Consequently, TOPSIS can be appropriately applied to estimate the composite scores (CS_i) for all options. In TOPSIS, the normalized decision values (n_ij) for each option x(i) in association with each minor criterion (j) can be calculated using Equation (7). Subsequently, the weighted normalized matrix (wn_ij) is computed by multiplying the normalized decision matrix (n_ij) by the relative weights (w_j) of each option corresponding each minor criterion. The P_j⁺ and N_j⁻ values for each minor criterion, estimated using Equations (8) and (9), are presented in

Table 11. An example of the calculations of the D_i⁺ and D_i⁻ values of the “Land use to support public transport (O16)” option.

Minor Criteria (Objectives)		Relative Weights (wj)	Normalized Values (nij)	Performance Scores (PSij)	Weighted Normalizeded Values (wnij)	Ideal Solutions		(wnij − Pj+)2	(wnij − Nj−)2
						(Pj+)	(Nj−)
Economic efficiency (C11)		0.194	0.184	0.8095	0.036	0.044	0.015	0.000071	0.000443
Economic growth (C12)		0.103	0.199	0.7143	0.020	0.023	0.012	0.000007	0.000067
Finance (C13)		0.078	0.244	0.7143	0.019	0.024	0.004	0.000026	0.000231
Livable streets and neighborhoods (C21)		0.074	0.228	0.9048	0.017	0.019	0.004	0.000003	0.000155
Equity and social inclusion (C22)		0.092	0.174	0.7143	0.016	0.022	0.007	0.000041	0.000073
Safety (C23)		0.201	0.199	0.8095	0.040	0.045	0.026	0.000022	0.000200
Environmental protection (C31)		0.258	0.226	0.9048	0.057	0.058	0.015	0.000000	0.001853
	Total							0.000170	0.003022
	Separation values							Di+ = 0.0131	Di− = 0.0550

, illustrating an example of the computations of the D_i⁺ and D_i⁻ values for the “Land use to support public transport (O16)” option. An illustration of the computation of the RC_i or CS_i value for the O16 option is = 0.0550/(0.131 + 0.0550) = 0.808. CS_i values achieved from TOPSIS were used to rank all 34 potential policy options.

The prioritization as well as the implementation status of these options for KKC are depicted in Figure 10. Based on the cluster analysis and careful consideration, the estimated composite scores (CS_i) of all 34 policy options, as shown in Figure 10, were classified into three distinct priority classes as follows: (1) high priority (CS_i ≥ 0.6000); (2) medium priority (0.5000 ≤ CS_i < 0.6000); and (3) low priority (CS_i < 0.5000).

5. Discussion

To generate, prioritize, and select a suitable set of policy measure options for SUPT planning in any medium-sized city (KKC) in any developing country (Thailand), the generation of all 34 potential options was achieved from the national SUPTM manual [20] for 78 regional cities of Thailand. Based on the HMADM approach used, the relative weights of the three main criteria and the seven minor criteria were determined using the FAHP for the city center area of a medium-sized city (KKC) in a developing country (Thailand). For the main criteria, the greatest relative weight was economic development (0.375), followed by social responsibility (0.367), and environmental protection (0.258). For the minor criteria, the highest relative weight was environmental protection (0.258), followed by safety (0.201), economic efficiency (0.194), economic growth (0.103), equity and social inclusion (0.092), finance (0.078), and livable streets and neighborhoods (0.074). As shown in Figure 7, the magnitude and the prioritization of the relative weights (in orange) of the main and minor criteria (objectives) are uniquely different from the relative weights (in gray) of the identical main and minor criteria achieved from the use of the AHP method for the small-town area in the same city (KKC) [48]. This reflects the fact that the relative weights of all main and minor criteria (objectives) potentially vary with the area types, even in identical cities.

In KonSULT KB [14], the contribution scores were generally obtained from human knowledge and expertise in European countries. The original 11-point scale ranging from −5 (strongest possible negative contribution) to +5 (the strongest possible positive contribution) was adopted in the original scoring system. These performance scores are employed to assess each option in achieving all seven minor criteria (objectives) containing unclear and vague (fuzzy) information. In addition, such original performance scores may lead to confusion and misinterpretation due to both positive and negative figures of the scoring system. The FSM was used to convert the linguistic (fuzzy) scores efficiently and systematically to their corresponding numerical (crisp) scores, as listed in Table 9.

Considering the estimations of the relative weights of the seven minor criteria (objectives) and the performance scores of all policy options with respect to the seven minor criteria, TOPSIS was used to determine the composite scores (CS_i) of all options, as shown in Figure 10. Noticeably, the prioritization of the 34 options for SUPT planning in the city center area of KKC, Thailand, is achievable. A total of 31 (91.2%) out of 34 policy options were recommended for KKC, which relied on those proposed options: 7 (22.6%) options were in the completely implemented (COI) stage, 7 (22.6%) options were in the being implemented (BI) stage, and 17 (54.8%) options were in the on-planning stage. Only 3 (8.8%) options out of 34 options were not suggested for KKC.

The top 13 options were categorized into the high-priority class. The ranking order of such top 13 options is as follows: (1) land use to support public transport (O16), (2) road user charging (O34), (3) pedestrian areas and routes (O6), (4) intelligent transport systems (O23), (5) development density and mix (O15), (6) fare structure (O30), (7) promotional activities (O1), (8) parking charges (O33), (9) bus rapid transit (O3), (10) integrated ticketing (O32), (11) fuel taxes (O31), (12) new rail stations and lines (O4), and (13) urban traffic control (O28). A total of 10 (77%) out of the top 13 options were proposed in the city’s SUPT plan for KKC. Only 2 (15.4%) options (O1 and O4) were in the CI stage, 3 (23.1%) options (O6, O23, and O28) were in the BI stage, 5 (38.5%) options (O16, O15, O30, O3, and O32) were in the OP stage, and finally all 3 (23.1%) options (O34, O33, and O31) were not proposed for KKC. The administrators and decision makers in both local and central government bodies are urged to critically focus on accelerating the implementation of those options in the BI and OP stages. Furthermore, the road user charging (O34), the parking charges (O33), and the fuel taxes (O31) options were not suggested for KKC due to the low political support and public opposition associated with them [13,19]. A similar interpretation can be applied to the remaining policy options in both the medium-priority and the low-priority classes. This means that the new DSF can be effectively applied to generate, rank, and select suitable SUPT policy measure options in medium-sized cities in developing countries. However, compared to all 34 options in the SUPTM manual [20], some policy options available in the KonSULT database [12] reveal their potential applicability to SUPT planning in any medium-sized city in developing countries. Some examples of these options are bike sharing, ride sharing, cycle networks, and demand-responsive transport. These policy options can be incorporated into the existing policy options of the SUPTM manual. In contrast, some examples of the KonSULT policy options, such as car clubs, new road construction, parking guidance systems, road freight fleet management, and lorry routes and bans, might not be practically relevant and suitable to SUPT planning in any medium-sized cities in developing countries. Furthermore, several policy options can be combined to encourage synergy and minimize operational barriers. In the future, the most appropriate prioritization and selection approach should also consider the packaging principle.

Moreover, public participation in SUMP planning on the objective’s definition level and the project level in developing (e.g., Southeast Asia) countries has been greatly inferior and restricted [19]. The new DSF can be applied to facilitate decision makers and other stakeholders in jointly determining potential options, suitable objectives, and the relevant performance scores of each option corresponding to all determined criteria (objectives). The HMADM component can then be used to assist decision makers and other stakeholders in cooperatively considering the relative weights of all decision criteria (objectives), and the prioritization of all options can finally be achieved. Hence, the DSF can be applied to enhance the participation of decision makers, various groups of stakeholders, and the public in generating, prioritizing, selecting, and implementing all determined policy options, as well as minimize the controversial aspirations and political barriers among them. Eventually, the DSF will be a powerful means to furnish relevant and useful information, increase the efficiency and success of public participation and consultations, allow making decisions together, and collectively implement appropriate policy measures for various groups of stakeholders and the public in the SUPT planning process of any medium-sized city in any developing nation [13,19].

6. Conclusions

The challenges of rapid economic growth, urbanization, and increased motorization in developing countries require a comprehensive and sustainable approach to the SUPT planning process, which has emerged as a critical solution to these complex challenges. However, the difficulty and complexity of generating, prioritizing, and selecting the appropriate policy measure options remain obstacles.

This study sought to develop a new DSF by harnessing the applicability of the national SUPTM manual of Thailand, the power of the KonSULT KB, and the robustness of the HMADM technique that integrates the FAHP, FSM, and TOPSIS. DSF is a rigorous mechanism for creating, prioritizing, and selecting SUPT policy options tailored to the unique context of medium-sized urban areas in developing countries. The FAHP was used to achieve expert-derived weights of the decision criteria, transform fuzzy scores into numerical values using the FSM, and finally rank policy options through TOPSIS. Collectively, this constitutes an innovative and comprehensive approach for addressing the intricate challenges of SUPT planning. The practical implementation of this framework was exemplified through a case study of the city center area (KKC), Khon Kaen, Thailand, which is a prime example of a medium-sized city in a developing country grappling with rapid urbanization.

The developed DSF proposes a set of 34 policy options with a specific ranking order for the KKC area. These options encompassed policy measures that focused on areas such as land use development, improvements to public transport, and the implementation of measures to restrain private vehicles while promoting public transport usage.

A comparative assessment was conducted between the proposed policy options and the actual options in the planning and implementation of KKC. The findings indicate that a significant proportion, namely 31 (91.2%) out of the total 34 options within the DSF, align with the city’s existing plan. These policy options are either fully implemented, in progress, or in the planning stages. However, of the top 13 options (categorized in the HP class) recommended by the DSF, 10 (77%) were incorporated into the city’s SUPT plan. Interestingly, only 2 (15.4%) policy options that were in the completely implemented (COI) stage did not rank as having the highest priority in the DSF recommendations. Notably, the city’s SUPT plan does not include 3 (23%) private vehicle restraint options (O34, O33, and O31). This omission may be attributed to political obstacles and a lack of public acceptance of such options.

Overall, the results demonstrate that the novel DSF framework may be more effective than the conventional SUPT planning process in the generation, prioritization, and selection of policy options to achieve the visions and objectives for a medium-sized urban area in a developing country. This can be used to guide administrators, decision makers, and strategic bodies in undertaking and adapting appropriate actions to implement SUPT policy measures. Additionally, the selected and prioritized options are beneficial for public participation activities (e.g., public hearings and consultations) involving various groups of stakeholders and the public in the planning process.

In practice, the proposed DSF suffers from several limitations and drawbacks. Some examples of such restrictions and disadvantages are as follows: (i) Based on the SUPTM manual [20], the list of 34 policy options determined in this research is far from complete. According to the context of urban areas, technological advancement, and the existence of other applicable options elsewhere, the expansion of new relevant and applicable options in the existing list of 34 potential options is strongly suggested. (ii) The performance scores of each option for all determined criteria (objectives) used in this research were obtained from the KonSULT KB previously generated for the developed (European) countries. Hence, the new performance scoring system specifically created for medium-sized cities in developing nations should be developed. (iii) Only two groups (both academia and decision makers) of 26 experts in urban land use and public transport planning and development, urban design, and town planning areas were chosen and interviewed directly to consider the relative weights of all criteria (objectives). All other stakeholders and the public were not included, and (iv) the prioritization of all 34 options in this research was only based on the HMADM component of the DSF without considering the packing principle.

Moreover, it is crucial for developing cities to seize the opportunity to measure and evaluate the effects of new measure options and share this valuable information with others. Insights into options that yield unexpected outcomes can play a pivotal role in helping others avoid similar pitfalls. Even if experience is available, direct applicability to different contexts may be limited, and assessing the transferability of experience for effective policy measures is challenging. This highlights the importance of encouraging documentation of as much experience as possible [8]. Additionally, the context in which measure options are implemented inevitably influences performance, which may not be readily transferable. Nevertheless, proper predictive models can mitigate these limitations by enabling the projection of impacts on demand and, consequently, on outcome indicators in various contexts [8].

The development and assessment of this DSF framework marks a significant advancement in the pursuit of SUPT planning. As developing countries continue to undergo transformative changes, the integration of knowledge, expertise, and advanced decision-making techniques will be pivotal role shaping efficient, equitable, and environmentally conscious urban mobility. As SUPT planning is one of the most important components of the SUMP, the SUPT planning of any city in any developing nation should expand its scope by taking the broader SUMP process into account in the future.