Integration of Principal Component Analysis with AHP-QFD for Improved Product Design Decision-Making

Abstract

The complexity of quality function deployment (QFD) matrices often hinders efficient decision-making in product design, leading to missed opportunities and extended development times. This study explores the integration of principal component analysis (PCA) with analytic hierarchy process-QFD (AHP-QFD) to address these challenges. PCA, a machine learning technique, was applied to QFD matrices from product design research to reduce complexity and enhance prioritization efficiency. The integrated method was tested with a product design team across various industries, including logistics, healthcare, and consumer electronics. The analysis demonstrated that PCA effectively reduced matrix complexity, optimizing feature prioritization. In the logistics sector, PCA explained 99.2% of the variance with the first five components, while in consumer electronics, it accounted for 86.9% with the first four components. However, PCA showed limitations in the healthcare sector due to evenly distributed variance among components. Expert feedback highlighted the practical benefits of the integrated approach: 75% of logistics experts and 62.5% of consumer electronics experts found the method clearer. For speed, 100% of logistics and 87.5% of consumer electronics experts preferred the method for quicker evaluations. For accuracy, 75% of logistics and 62.5% of consumer electronics experts deemed the method more accurate. Overall, the PCA-AHP-QFD method simplifies decision-making processes and reduces development time, particularly in industries where feature prioritization is crucial. These findings underscore the potential of the integrated approach to enhance product development efficiency and feature prioritization, with suitability varying based on industry characteristics.

1. Introduction

The current wave of digitization has profoundly transformed the world, significantly impacting product design. Understanding customer needs and addressing their requirements are crucial factors influencing market success. Additionally, quality requirements and cost-efficient management remain critical for products to be competitive and sustainable [1,2,3]. These factors continue to be vital considerations [4,5,6,7].

Recent trends show increasing product requirements and rapidly evolving markets demanding highly precise product developments in shorter cycles. Production dependencies and boundary conditions must be considered early in the product design phase to ensure functionality [8,9]. This heightened complexity necessitates precise designs with numerous variables, making design decisions more challenging and extending development times [8,9,10,11].

Advancements in information technologies offer significant opportunities to support the product engineering process through increased computing power, new simulation and analysis tools, and interconnected data [10]. Digital twins of products or production processes are now being modeled in various domains to derive optimal solutions [8]. Enhanced data availability and traceability facilitate the creation of data-driven models using artificial intelligence (AI), big data analytics, and machine learning (ML) methods [10,12,13]. These advancements enable the development of algorithms that support or even substitute human decision-making [14,15]. Data mining techniques uncover patterns in big data, enabling the discovery of unexpected knowledge without prior theorization and domain expertise [16,17].

Despite these benefits, incorporating digital technologies into product design presents notable challenges. Automation can erode tacit knowledge traditionally held by designers, complicating the effective use of human expertise [8,10,11,18,19]. Effective implementation requires collaboration among product designers, production planners, and application engineers, each with unique boundary conditions and optimization strategies that can unpredictably affect subsequent phases [8]. Clearly defining roles and scopes of action is crucial for achieving optimal solutions.

While supplementary reviews from online sources provide valuable insights, their impact on prioritizing customer requirements has not been extensively examined [20]. In group decision-making, opinion inconsistencies often arise, making consensus essential [20]. Addressing these issues requires advanced technological tools and effective communication and collaboration among stakeholders.

Traditional methods, such as multi-criteria decision-making (MCDM) techniques like quality function deployment (QFD) and analytic hierarchy process (AHP), continue to be studied and developed to assist in product design. These methods will be discussed further in the literature review section. As shown in

Table 1. AHP-QFD approaches in product design decision-making.

Study References	Approach	Key Benefits	Industrial/Area
[23]	AHP-QFD combined with zero-one goal Programming (ZOGP) binary model	Effective site selection for cargo logistics centers, managed budget constraints, demonstrated managerial decision-making effectiveness	Cargo Logistics Centers
[24]	AHP-QFD method	Enhances prioritization of customer requirements in new product development, leading to better design alternatives.	New Product Development
[25]	QFD integrating Kano model and fuzzy AHP	Improves value in product–service systems (PSSs) by focusing on customer requirements and managing uncertainty.	PSS
[26]	QFD and AHP	Improved design quality in buildings, better selection of design criteria, increased customer satisfaction	Building Design
[27]	AHP-QFD	Enhanced student information system, alignment of improvements with institutional capabilities	Educational Systems
[28]	AHP-QFD	Balanced cultural preservation with commercial value, innovative renovation method for traditional houses	Traditional Residential Renovation
[29]	AHP-QFD	Comprehensive framework for smart parking system design, enhanced user satisfaction and operational efficiency	Smart Parking Systems
[30]	AHP-QFD	Ensures that the new design meets stakeholder requirements, improving the usability and functionality of temperature-controlled medicine bags.	Medical Equipment Design

, MCDM methods such as AHP and QFD are widely used in combination for decision-making in product design. Summarizing the key strengths and limitations of MCDM combined with QFD, such as AHP, analytical network process (ANP), and fuzzy logic [21,22], it is evident that no single method can address all key challenges comprehensively.

Therefore, this study focuses on combining AHP and QFD to improve product design decision-making, leveraging their individual strengths for a more structured and quantitative approach. However, the complexity of the AHP-QFD approach remains a significant challenge. Techniques for reducing complexity or dimensionality, such as principal component analysis (PCA) [31,32], are incorporated to facilitate easier analysis and decision-making.

From the literature review on the PCA-QFD approach, it is evident that while PCA-QFD has potential, it remains unsuitable for some applications. The methodology itself may be correct, but input data likely play a crucial role, suggesting that resolving data issues could improve its effectiveness. This consideration leads us to enhance the PCA-QFD approach by integrating AHP, motivating the study of the PCA-AHP-QFD methodology.

The combined use of PCA and AHP offers significant advantages. PCA reduces data complexity, making it more manageable, while AHP provides a structured decision-making framework. Studies have highlighted the benefits of using PCA and AHP together to enhance decision-making accuracy and efficiency [33,34].

By integrating QFD, AHP, and PCA, the strengths of each method are leveraged to comprehensively address the limitations of QFD. This integrated approach ensures that customer needs are prioritized objectively, critical components are identified efficiently, and data complexity is reduced, leading to a more effective and streamlined product development process. Although previous studies have shown the benefits of using PCA or AHP with QFD, there has been limited research on combining all three methods. This study aims to fill this gap by evaluating the applicability and benefits of this integrated approach.

In particular, this study addresses several key challenges as follows: the increasing complexity of designs extending development times, the demand for advanced skills complicating management, the loss of tacit knowledge, the need for collaboration among stakeholders with differing goals, and achieving consensus in group decision-making. By Cincorporating PCA with MCDM methods like AHP and QFD, this study aims to simplify data analysis, enhance decision-making accuracy, and comprehensively address these challenges. Details about PCA will be provided later in the literature review, specifically in Section 2.

2. Literature Review

2.1. AHP and QFD Approach in Product Design Decision Making

Designing new products often presents significant challenges, particularly in ensuring that customer needs are at the forefront of the design process. This approach is essential because matching the needs and various characteristics for product design is the key point to develop a successful product [35,36]. Effectively addressing customer requirement can provide a substantial market advantage, as products that closely align with customer desires are more likely to succeed [36,37,38]. In the competitive landscape, being able to respond to customer needs efficiently and accurately is a key differentiator.

AHP aids in prioritizing customer requirements by structuring them into a hierarchy and performing pairwise comparisons to assign relative weights. This quantitative approach reduces subjectivity and enhances the accuracy of prioritizing customer needs [39,40,41]. QFD, on the other hand, translates these prioritized requirements into specific engineering characteristics, ensuring that customer needs are systematically addressed throughout the product development process [1,42].

In the competitive landscape, being able to respond to customer needs efficiently and accurately is a key differentiator. The combined use of AHP and QFD helps in managing the complexity of product design by breaking down decision criteria into more manageable components and ensuring that all customer requirements are considered [43,44].

These approaches are detailed in Table 1, which presents a comprehensive overview of the application of the AHP-QFD approach across various domains of product design decision-making. The table highlights key benefits derived from integrating AHP-QFD and provides relevant references for each study.

Table 1 highlights the methodological innovations and key benefits derived from this integration. Each entry in the table provides insights into how AHP-QFD has been utilized to address specific decision-making challenges, improve prioritization of requirements, and enhance overall design outcomes.

2.2. Comparative Analysis of Key Strengths and Limitations of MCDM Methods Combined with QFD

In Table 1, it is clearly demonstrated that the AHP-QFD method is a widely used and continuously applied MCDM approach. However, other MCDM methods are also being utilized and investigated to further refine QFD techniques. Thus, analyzing the key strengths and limitations of various MCDM methods employed in design decision-making can enhance our understanding and lead to improvements in these methodologies.

Table 2. Key strengths and limitations of MCDM methods combined with QFD.

Study Reference	QFD	AHP AHP-QFD	Fuzzy Logic, Fuzzy QFD, Fuzzy AHP-QFD	ANP ANP-QFD
Strength
[45]	This study does not provide information about this part.	AHP is simple and easy to understand, making it accessible to decision-makers without a strong background in mathematics. It is effective for multi-criteria decision making and provides a clear hierarchical structure.	Fuzzy methods are effective in managing uncertainty and imprecision in judgments. They handle inconsistent data well and provide a flexible decision-making framework.	ANP handles interdependence among criteria and feedback effects, making it more flexible and realistic in complex decision-making scenarios, and providing a more comprehensive analysis than AHP
[46]	QFD is commonly used in the field of new product development to translate customer requirements into appropriate engineering characteristics. It demonstrates how quality characteristics can effectively satisfy customer requirements.	QFD-AHP enhances the effectiveness of the decision-making process and is also utilized to manage the subjective linguistic judgments that arise when expressing relationships and correlations necessary in the QFD approach.	Fuzzy logic has been adapted into QFD to minimize the vagueness frequently present in decision data.	This study does not provide information about this part.
[47]	A simplified form of QFD in product design offers an easy method for implementation, analysis, and documentation. It effectively connects customer needs with product features, enhances teamwork communication, and reduces product development cycle times.	These methods, discussed in terms of comprehensive or improved versions of QFD, offer opportunities to gain a competitive advantage by better meeting client needs. These versions emphasize teamwork and improve the systematic nature and prioritization of client requirements, notably reducing the subjective nature typically associated with the initial phases or simplified versions of QFD.
[25]	QFD is highly effective in capturing and translating customer needs into specific engineering characteristics. It provides a systematic process for prioritizing customer requirements and enhances communication among different departments.	AHP-QFD integrates hierarchical structuring and prioritization capabilities, allowing for quantitative comparison and improving decision-making accuracy. This combined approach results in a clearer and more structured decision-making process.	Fuzzy QFD uses fuzzy logic to manage ambiguity in customer requirements, reduces biases by converting subjective evaluations into fuzzy numbers, and improves decision-making with uncertain information.	The QFD for the PSS method with ANP to examine the relationships between the ‘whats’ and ‘hows’ of each matrix could be beneficial. It may enhance the understanding of the inter- and intra-relationships within PSS.
Limitations	QFD	AHP AHP-QFD	Fuzzy logic, fuzzy QFD, fuzzy AHP-QFD	ANP ANP-QFD
[45]	This study does not provide information about this part.	AHP assumes independence among criteria and is limited in handling interdependence and feedback.	Fuzzy logic can be computationally intensive and requires significant resources and time. Its accuracy depends heavily on the precise definition of membership functions, which can be challenging to determine.	ANP is more complex and time-consuming due to its consideration among criteria. And requires detailed pairwise comparisons.
[46]	QFD has limitations due to subjective linguistic judgments when expressing relationships and correlations required in the House of Quality (HOQ). Additionally, it often deals with vagueness in decision data.	The main limitations of AHP highlighted are the sensitivity to the addition of alternatives and the need for more specific pairwise comparison questions to ensure rank preservation and consistency.	This study does not provide information about this part.	This study does not provide information about this part.
[47]	The simplified form of QFD, its primarily qualitative and subjective nature, which can make categorizing the fuzzy nature of customer needs challenging. Additionally, it may not be suitable for all industries.	These methods in terms of comprehensive or improved versions of QFD, particularly when using more complex, multi-matrix approaches. The size of all matrices and the analysis can become very large, leading to challenges in implementing numerous conclusions derived from the analysis. It also demands extensive knowledge and skills from the team executing the method. If the team lacks sufficient understanding or skills, it can further slow down the performance of QFD analysis.
[25]	QFD process often involves subjective judgments, which can lead to biases. Additionally, maintaining and updating the QFD matrices can be challenging).	The integration process can be resource-intensive, requiring detailed data collection and extensive pairwise comparisons. It can also be complex to implement, especially for teams unfamiliar with both AHP and QFD methodologies. Ensuring consistency in pairwise comparisons can be challenging	Implementing fuzzy logic is computationally intensive and requires significant resources and time. Its accuracy depends on precise membership function definitions, which can be challenging, and there is a steep learning curve	This study does not provide information about this part.
[48]	QFD does not clearly minimize customer requirements and feature requirements	This study does not provide information about this part.	The use of ANP and fuzzy QFD is difficult for obtaining the correct network structure and minimizing requirements from QFD, especially when compared to unsupervised machine learning techniques.

offers a comprehensive comparative analysis of the key strengths and limitations of these methods. By examining various MCDM techniques which usually combine with QFD, such as AHP, fuzzy AHP, and ANP, a more nuanced understanding of their effectiveness, challenges, and applicability across different contexts can be attained.

Based on Table 2, it can be concluded that each process has its own strengths and limitations, which can be improved and tailored to the QFD approach depending on the context and objectives of the application.

QFD is highly effective in translating customer requirements into engineering characteristics, making it widely used in new product development due to its systematic process, ease of implementation, and ability to enhance communication and reduce development cycles. However, its reliance on subjective judgments can introduce biases and vagueness, and its qualitative nature may pose challenges in categorizing customer needs, making it less suitable for certain industries. Additionally, maintaining and updating QFD matrices can be difficult, complicating the decision-making process.

Among MCDM, it is evident that each method has distinct strengths and limitations that can be tailored and improved upon to suit the QFD approach, depending on the specific context and objectives. For this study, the aim is to address key challenges in product design decision-making.

The primary advantage of AHP is its ability to provide a quantitative method to prioritize customer requirements and technical characteristics based on their relative importance. This method adds objectivity and rigor to the decision-making process, reducing the subjectivity and bias inherent in QFD.

Compared to other methods like ANP and fuzzy logic, ANP also handles complex decision-making processes [25,26,45,47,49]. It is also complex and time-consuming due to its consideration of interdependence among criteria and detailed pairwise comparisons [45,47,48] Fuzzy logic, on the other hand, deals well with uncertainty and imprecision [25,45,46,47,50], but also can introduce additional complexity and be difficult to use [25,45,47,48]. These methods demand extensive knowledge and skills, and if the team lacks expertise, it can slow down the QFD analysis. AHP simplifies the process by breaking down complex decisions into a hierarchy of more easily comprehended sub-problems, each of which can be improve independently [25,45,46,47]

Thus, AHP-QFD is chosen for improvement due to its combined strengths and easier implementation compared to ANP and fuzzy logic, which, while having significant advantages, require advanced skills and are time-consuming to analyze and understand.

2.3. Integration of PCA to Address QFD Limitation: The Complexity of Matrices

While the integration of AHP with QFD helps address the subjectivity of the ratings used, the complexity of the QFD matrices remains a limitation. To address the limitations of AHP-QFD without encountering the difficulty to use method such as ANP and fuzzy logic [48], PCA is proposed.

2.3.1. PCA and Its Role in QFD Context

PCA, introduced in the early 1900s and developed further in the 1930s, is a statistical method that reduces the dimensionality of datasets while retaining most of the original variability. It simplifies complex datasets by transforming the original variables into a new set of uncorrelated variables called principal components, ordered by the variance they explain [31,32].

The primary advantage of PCA is its ability to reduce the complexity of large datasets [51,52], making it easier to analyze and interpret compared to other dimensionality reduction techniques like independent component analysis (ICA) and factor analysis (FA) [48]. In the context of QFD, for PCA, the primary goal is often to simplify and prioritize customer requirements and product features while retaining the most significant variance in the data. PCA is particularly effective in this scenario.

In contrast, the study on spectral-temporal data found ICA and FA to be more effective because these methods are designed to handle specific types of data distortions and preserve the integrity of important components in ways PCA cannot. The different nature of the data and the distinct goals of the analysis in that study explain why ICA and FA outperform PCA in that specific context [53]. Therefore, while ICA and FA might be superior for handling highly distorted data in some contexts, PCA’s simplicity, efficiency, and alignment with the goals of QFD make it a preferred choice for dimensionality reduction in QFD applications.

By identifying and focusing on the most important components, PCA helps streamline the QFD process, enhancing its efficiency and effectiveness. This is particularly useful in handling the complexity of QFD matrices, where multiple customer requirements and technical characteristics need to be considered simultaneously [48,54].

The integration of PCA with QFD has been applied to address the complexity of QFD matrices. For example, since 2007, PCA-QFD integration has been successfully utilized to handle sparse high-dimensional data, demonstrating PCA’s effectiveness in managing complex datasets and improving decision-making in product development [55]. More recently, PCA was applied to optimize the voice of stakeholders in the QFD method, streamlining stakeholder inputs and enhancing decision-making in the development process [54].

However, not all applications of PCA-QFD integration have been successful. For instance, an attempt to use PCA-QFD integration in selecting functional requirements in construction automation was found to be inappropriate. This was due to the specific requirements and constraints of the construction industry, which rendered the PCA approach less effective in this context [48].

2.3.2. Integration of PCA with QFD

PCA, involving a reconstructed dataset $\widehat{\mathrm{X}}$ approximating X through a scores matrix (T) and a loadings matrix (P), was utilized in this study for unsupervised feature selection to determine the importance of feature requirements in meeting customer expectations as the algorithm represented in Figure 1. It outlines the algorithm of feature selection using PCA for feature requirements.

The algorithm begins by normalizing the dataset X to have zero mean and unit variance, which is crucial for PCA analysis. Next, the covariance matrix C_x is computed using the normalized dataset X, given by Equation (1), where m is the number of samples and X^T is the transpose of X.

$$C_x=\frac{1}{m-1}X{X}^T$$

(1)

Following this, the eigenvectors (p_i) and eigenvalues (λi) of the covariance matrix Cx are calculated by Equation (2).

$$C_x{p}_i={\lambda }_i{p}_i$$

(2)

The eigenvectors represent the principal components, and the eigenvalues indicate the amount of variance explained by each principal component. The eigenvectors are then sorted in descending order based on their corresponding eigenvalues to identify the most significant principal components that capture the maximum variance in the data, or utilizing singular value decomposition (SVD) to automatically rank principal components (PCs) based on their singular values following Equation (3).

$$\mathrm{X}=\mathrm{U}\mathsf{\Sigma }{\mathrm{V}}^T$$

(3)

The delineation of the SVD of a matrix (3) serves as a pivotal tool for dimensionality reduction in the context of this algorithm. It elucidates the decomposition of the original matrix X into its constituent components as follows: U, the matrix of left singular vectors; Σ, the diagonal matrix of singular values; and either V or its transpose τ, representing the right singular vectors matrix.

Then, the loadings matrix (P) is determined by minimizing the reconstruction error, and the expected value of the reconstruction error is calculated as the sum of minor eigenvalues [48].

$$P={\mathrm{a}\mathrm{r}\mathrm{g}\mathrm{m}\mathrm{i}\mathrm{n}\parallel X-XP{P}^T\parallel }^2$$

(4)

Equation (4) indicates that P = V since PP^T is the orthogonal projection of X onto the subspace spanned by the columns of V. Thus, the loadings, scores, and reconstructed dataset are derived through the singular value decomposition of X following Equation (5).

$$P=V$$

(5)

The algorithm involves steps such as normalizing the dataset, calculating the covariance matrix, computing the eigenvectors and eigenvalues, sorting the eigenvectors in descending order of eigenvalues, and selecting the top eigenvectors as principal components. By selecting the principal components with the largest eigenvalues, the algorithm aims to capture the maximum variance in the data, thus identifying the most critical FRs for meeting customer requirements. The algorithm’s objective is to reduce the dimensionality of the dataset while retaining as much variance as possible, simplifying the feature requirements selection process in QFD matrices. Through PCA, the algorithm provides a systematic approach to identifying the most influential feature requirements, facilitating efficient decision-making in the conceptual design phase of product and service.

3. Materials and Methods

In the initial phase, AHP-QFD matrices from the product design phase across various industrial research fields were gathered. These matrices were subjected to PCA to assess the fitness of the PCA method for each matrix. Only the matrices deemed appropriate based on the fitness model were selected for further evaluation.

Subsequently, PCA was implemented on the selected AHP-QFD matrices using the Python programming language. The results obtained from both the original QFD matrices and the PCA-transformed QFD matrices were then assessed by product designers to evaluate their interpretability and usability in decision-making processes related to product and service design.

Finally, the insights gained from the comparison were synthesized to determine whether PCA can enhance the AHP-QFD matrix in feature selection optimization and whether the effectiveness varies across different industry sectors.

3.1. Research Framework

This study addresses the need to assess the suitability of PCA in reducing the complexity of AHP-QFD matrices during the product design phase, aiming to refine decision-making processes and optimize feature selection.

Its primary objective is to evaluate the fitness or suitability of the PCA model with the AHP-QFD matrix data, while also discerning the impact of PCA on feature selection optimization and variations in effectiveness across divergent industrial sectors.

The components of the framework include original AHP-QFD matrices sourced from various industrial research fields, the application of PCA to reduce complexity and identify salient features, the evaluation of the PCA model’s performance with the AHP-QFD matrix data through a fitness assessment, the evaluative scrutiny of both original and PCA-transformed QFD matrices by product designers to assess interpretability and usability, and the synthesis and analysis of insights gleaned to discern the impact of PCA on feature selection optimization and identify sectoral disparities. The overall structure is shown in Figure 2.

3.2. Initial AHP-QFD Matrices

The selection of the three case studies by Tu et al. [23], Buakum et al. [30], and Tu et al. [24], for the initial AHP-QFD matrices to integrate with PCA as PCA-AHP-QFD approach is based on several key reasons. These reasons align with the key challenges faced in product design decision-making, such as the need to reduce development time, achieve sustainable product design, and ensure ease of understanding for all stakeholders involved, and still align with customer or user requirements. The chosen case studies span different application domains as follows: airport ground handling services, medical equipment design, and consumer electronics. This diversity ensures that the PCA-AHP-QFD approach is tested across varied contexts, enhancing its generalizability and applicability to different industries.

Each case study addresses multifaceted decision-making scenarios. Tu et al. [23], dealt with the selection of optimal locations for cargo logistics centers, requiring consideration of multiple criteria and budget constraints. Buakum et al. [30], focused on designing a product that met the needs of multiple stakeholders in the healthcare sector. Tu et al. [24], highlighted the importance of prioritizing customer requirements in new product development. The complexity and depth of these cases provide a robust foundation for testing the effectiveness of integrating PCA with AHP-QFD.

The selected studies employed rigorous methodological approaches that combine AHP with QFD effectively. Tu et al. [23], incorporated zero-one goal programming to manage budget constraints, while Buakum et al. [30] and Tu et al. [24], utilized comprehensive data collection methods (questionnaires and interviews) to capture customer needs and preferences. This methodological rigor ensures that the initial AHP-QFD matrices are well-founded, providing a solid basis for the integration of PCA.

All three case studies emphasize the importance of capturing and prioritizing stakeholder preferences. By considering the professional experiences of QFD teams, gathering inputs from pharmacists, patients, and their relatives, and focusing on customer requirements for new product development, this focus on stakeholder preferences aligns well with the objectives of integrating PCA to streamline and simplify the decision-making process.

The selected case studies are categorized by industry as follows.

3.2.1. Logistic Service Site Design

The study by Tu et al. [23], focused on selecting optimal locations for cargo logistics centers, addressing multifaceted decision-making with budget constraints. The illustration of the Airport Cargo Logistic Center Site is presented in Figure 3. This illustration provides an overview of the example key components of a logistic center. The main elements include airport land features, operating handling costs, carrying conditions, and proximity to airline agents and customers. Smaller subcomponents highlight aspects such as local political regulation and law, improvement district’s working, airport labor status, electrical power/utilities, and information technology status. This visual representation helps in understanding the overall design and functionality of the logistic center.

These components will be analyzed as quality requirements from stakeholders, and the analysis will be conducted through AHP-QFD. The results will be presented in the structure shown in Figure 4 to aid in product design decision-making and the initial data from the analysis are displayed in

Table 3. Original AHP-QFD matrix of site selection for an Airport Cargo Logistics Center [23].

Quality Requirement	Airport Cargo Logistic Center Site Requirement									Important Weighting of Requirements
	FR1: Airport Land Feature	FR2: Considering Airport Operating Handling Cost	FR3: Airport Carrying Condition	FR4: Closeness to Airline Agent and Customer	FR5: Local Political Regulation and Law	FR6: Improvement District’s Working	FR7: Airport Labor Status	FR8: Airport Electrical Power/Utilities	FR9: Airport Information Technology Status
CR1	3	9	9	3	9		9	3	9	0.300
CR2	9	9	9	3	3	3	3		9	0.060
CR3	9	9		3	9	3		1		0.087
CR4	9	9	3	3					3	0.030
CR5		9		3					3	0.127
CR6	3	9		3		9				0.124
CR7		3		3		9	9			0.042
CR8				9	1	3	3		9	0.021
CR9						9	9	9		0.017
CR10		9		9		3		9	9	0.192
Importance degree	2.865	8.406	3.330	4.227	3.687	2.727	3.474	2.868	5.628
Normalize important degree	7.7	22.6	8.9	11.4	9.9	7.3	9.3	7.7	15.1

Note: The list of customer requirements associated with this AHP-QFD include: (CR1) quick/accurate logistic ability, (CR2) comfortable logistics, (CR3) adequate building space, (CR4) proper logistics center site, (CR5) reasonable logistics fee, (CR6) easy to load/unload, (CR7) adequate airport service supply, (CR8) sufficient public facilities, (CR9) stable economic utilities provide, (CR10) quick delivery service capability.

.

Components of the Matrix:

• Airport Cargo Logistic Center Site Criteria (quality characteristics): At the top of the matrix, lists the criteria that are considered essential for selecting the optimal site for the cargo logistics center.
• Airport Cargo Logistic Center Site Requirements (quality requirements): On the left side of the matrix, this section outlines the specific requirements that the logistics center must meet. These requirements are derived from the needs of stakeholders such as logistics companies, airport authorities, and customers.
• Central Relationship Matrix: The central portion of the matrix shows the relationships between the site criteria and the site requirements. Each cell within this matrix indicates the degree of correlation or impact that a particular criterion has on fulfilling a specific requirement. The strength of these relationships is typically represented using symbols or numerical values (e.g., strong (9), moderate (3), weak (1)).
• Importance Weighting of Requirements: On the right side of the matrix, the importance weighting of each requirement is indicated. The AHP is used to measure the relative importance of the weights for each site requirement.
• Importance Degree of Airport Cargo Logistic Center Site Criterion: At the bottom of the matrix, this row displays the importance degree of each site criterion based on its ability to meet the defined requirements. This helps prioritize the criteria that are most critical for the site selection process.
• Normalized Importance Degree of Airport Cargo Logistic Center Site Criterion: The final row shows the normalized importance degree of each site criterion. Normalization adjusts the importance degrees so that they sum up to a specific value.

Based on Table 3, the AHP-QFD matrix supports decision-making in selecting suitable site attributes. This study interprets the matrix results, but, as mentioned earlier, the interpretation depends on the skills and expertise of the decision-makers.

In this study, the important degrees were considered in the following order of priority: the airport’s operating handling costs (8.406) were deemed the most important, followed by the airport’s information technology status (5.628). Next in importance was the closeness to airline agents and customers (4.227), then the local political regulations and laws (3.684). The airport’s labor status (3.474) was next, followed by the airport’s carrying condition (3.330), the airport’s electrical power/utilities (2.868), and the airport’s location conditions (2.865). Finally, the improvement district’s working environment (2.727) was considered the least important among the listed factors. However, this case study indicates that results cannot be determined solely from the AHP-QFD matrix. Other techniques, such as zero-one goal programming, are needed to facilitate decision-making. It is recommended that future studies apply other techniques in conjunction with AHP-QFD to potentially improve decision-making outcomes.

3.2.2. Medical Product Design

The study by Buakum et al. [30], focused on designing a temperature-controlled medicine bag as shown in prototype components in Figure 5. To address the shortcomings of the current product used at Songklanagarind Hospital in Thailand, the integration of the AHP and QFD techniques was employed to ensure that the new design met the requirements of various stakeholders, including pharmacists, patients, and their relatives, through the research framework as shown in Figure 6.

Figure 5 provides a comprehensive view of the prototype components of the temperature-controlled medicine bag. The outer bag ensures insulation, the inner bag provides the primary storage area, the cooling agent bag helps maintain the temperature, and the temperature indicator label allows for easy monitoring.

Figure 6 illustrates the framework used to identify key technical specifications for the temperature-controlled medicine bag through the integration of AHP and QFD techniques. The process consists of three main stages as follows: customer need analysis, AHP-QFD technique implementation, and product design.

The process of developing the temperature-controlled medicine bag prototype began with the customer need analysis. The objective was to gather and assess the needs and requirements of customers. The activities involved identifying customer requirements, evaluating the importance and priority of these needs, and assigning importance ratings to different customer groups based on their specific requirements and preferences. The output was a comprehensive understanding of customer needs, which informed the subsequent stages of the design process.

Next, the AHP-QFD technique implementation aimed to prioritize the important needs identified in the previous stage and develop a structured plan for addressing them. This stage involved using the AHP to rank the importance of the identified needs and creating a House of Quality matrix to translate customer needs into technical specifications that the product must fulfill. The output was a prioritized list of customer needs and a detailed matrix that served as a blueprint for product design.

Finally, the Product Design phase focused on developing and testing the product prototype based on the specifications derived from the matrix. Activities included designing and creating a prototype of the temperature-controlled medicine bag and conducting tests to ensure that the prototype met the technical specifications and customer requirements.

The initial AHP-QFD data from this case study are represented in the same QFD matrix structure as shown in Figure 2 for the logistics cargo center case, but applied to the context of the temperature-controlled medicine bag design. The detailed analysis is presented in

Table 4. Original matrix from AHP-QFD of a controlled-temperature medicine bag design ([30].

	Controlled Temperature Medicine Bag Design Features														Important Weighting of Requirements
* Customer Requirements	FR1: Insulation Ability of Inner Layer	FR2: Heat Reflection Ability	FR3: Anti-Shock Ability	FR4: Cooling Ability	FR5: Temperature Indicator	FR6: Ziploc Enclosure	FR7: Carry Handle	FR8: Bag Form	FR9: Weight	FR10: Size	FR11: Medical Disposable Item	FR12: Reusable	FR13: Safety Material	FR14: Price
CR1	9	9		9	1	3		1				1			4.94
CR2	9	9		9	1	3						3			4.91
CR3					9										4.66
CR4			9	1				3				1	9		4.96
CR5	3	9				3		1					1		4.84
CR6	3	3	3	3				9	9	9				3	4.43
CR7			3			9	3	9	9	9	9		3	3	3.71
CR8	3	3	3	1	9	3	1	9			9		9		4.87
CR9	1	1	1	1				3	9	9	3			3	3.82
CR10						9		3				1	9		4.98
CR11								3		9					4.69
CR12	1	1	9	1	9	9	1	3			9		9		5.00
CR13							9	3	3	3					3.54
CR14						9	3	9			9		3		5.00
CR15								3	3	3	9	1		9	4.83
CR16	9	9	9	9	9	1	1	9			3		9		5.00
CR17	3	3	3	3	3	1	1	3				9			3.35
TRs weight	194.94	223.98	187.54	175.64	195.67	235.24	76.21	322.38	132.75	174.96	237.15	64.30	254.26	79.35
Important (%)	7.63	8.77	7.34	6.88	7.66	9.21	2.98	12.62	5.20	6.85	9.28	2.52	9.95	3.11

* The list of customer requirements associated with this AHP-QFD include: (CR1) keep cool, (CR2) cooling generator, (CR3) temperature display, (CR4) anti-shock, (CR5) external heat protector, (CR6) temperature display, (CR7) bag form, (CR8) trustworthy, (CR9) lightweight, (CR10) covering, (CR11) sufficient capacity, (CR12) safety, (CR13) easy to carry, (CR14) easy to clean, (CR15) suitable price, (CR16) quality material, and (CR17) other utilities.

.

Table 4 presents the AHP-QFD matrix used to transform customer needs into important product control characteristics for designing a prototype of a temperature-controlled medicine container. The dominant features of the product prototype include a “light and small bag with Ziploc enclosure,” “made from safe and disposable medical materials”, “comes with a cooling agent and temperature indicator”, and “lower price than the market”, with relative importance values of 12.62%, 9.95%, 9.28%, and 9.21%, respectively.

The AHP-QFD matrix provides a structured approach to prioritize customer needs and align them with technical specifications. However, additional brainstorming was necessary to determine the relative technical requirement importance (%), using thresholds of 9% and 7%, which requires significant skill. From this process, eight technical requirements (TRs) were selected: TR8 (12.62%), TR13 (9.95%), TR11 (9.28%), TR6 (9.21%), TR2 (8.77%), TR5 (7.66%), TR3 (7.34%), and TR1 (7.63%).

In conclusion, while the AHP-QFD matrix is effective, it requires further refinement through brainstorming to ensure accurate decision-making. Without this additional step and the application of relative importance thresholds, making informed decisions based solely on the matrix and requirement weights would be challenging.

3.2.3. New Personal Electronic Product Development Design

The study by Tu et al. [24], focused on the application of a combined AHP with the traditional QFD method in the development of new products, specifically in a new earphone development project in Figure 7. The limitations addressed by integrating AHP with QFD in the study were as follows: (1) lack of prioritization: traditional QFD does not prioritize customer requirements and (2) subjectivity: the weights assigned to customer requirements are based on subjective evaluations and depend on the consensus of a panel of experts.

The process begins with data collection through questionnaires, interviews, and observations to identify customer requirements. These requirements are then analyzed and prioritized using AHP, which provides a structured method for determining their importance. The prioritized needs are translated into technical specifications using the House of Quality matrix structure. Finally, a prototype is developed and tested to ensure it meets the specified requirements, ensuring that the product aligns with customer needs and preferences.

The House of Quality matrix structure follows the same QFD structure as explained in Figure 1, but assigned a numeric value; strong (3), moderate (2), weak (1) excluded benchmarking section. The initial AHP-QFD data will be presented in the same way as in the previous two studies. The initial data for this case are shown in

Table 5. Original matrix from AHP-QFD of new earphone development [24].

	New Earphone Development Technical Requirements										Important Weighting of Requirements
* Customer Requirements	FR1: Material	FR2: Structure	FR 3: Wireless	FR4: Sealing	FR5: Battery Capacity	FR6: Power Consumption	FR7: Appearance	FR8: Weight	FR9: Connection Wire Length	FR10: Diaphragm Design
CR1	2				3			3			0.239
CR2	2	3									0.143
CR3			3						3		0.022
CR4		1								3	0.058
CR5										3	0.057
CR6	2	2		3							0.026
CR7	2	3						1	2		0.099
CR8		3						2	2		0.037
CR9					3	2					0.077
CR10	2	2					3				0.242
FRs Weight	196	179	5.4	17	171.7	11.37	60	182	59.74	121.1

* The list of customer requirements associated with this AHP-QFD include: (CR1) lightweight, (CR2) hold position, (CR3) wireless, (CR4) open ear for health, (CR5) open ear for safety, (CR6) waterproof, (CR7) comfort, (CR8) easy to carry, (CR9) continuous usage, (CR10) cool appearance.

.

Table 5 presents the AHP-QFD matrix for the development of new earphones. This matrix, resulting in higher accuracy than the commonly used scale point evaluations. For example, customer requirements like light weight (CR1) and hold position (CR2), which were both assigned a value of 5 in point scales and could not be differentiated, were given precise weights of 0.239 and 0.143, respectively, using AHP. The matrix identified key technical requirements such as material (FR1) and weight (FR8) with weights of 196 and 182, respectively, indicating their critical importance in the design process.

While AHP helps to prioritize customer requirements more accurately than simple point scales, it still involves subjective pair-wise comparisons. This subjectivity means that the values assigned can vary based on the judgment of the decision-makers. Therefore, if accuracy is paramount, AHP is preferred despite its subjective nature. However, for considerations of time, cost, and simplicity, the QFD method alone might suffice.

3.3. Apply PCA with Derived AHP-QFD Matrices

Based on the three case studies, the AHP-QFD method effectively supports decision-making by prioritizing customer needs and aligning them with technical requirements. However, its reliance on subjective pair-wise comparisons and the complexity of interpreting results indicates a need for further refinement. The potential of this study suggests that integrating PCA can address these limitations by simplifying the data and reducing redundancy.

In this study, the AHP-QFD matrices extracted from preceding investigations (Table 3, Table 4 and Table 5), as illustrated in Table 6, Table 7 and Table 8, are leveraged. These tables serve as the primary datasets, referred to as Dataset Xi, for the application of the PCA method. Within these matrices lies a comprehensive amalgamation of customer requirements into the design processes of products and services across various industrial domains, alongside the AHP-derived significance weights assigned to each customer requirement.

3.3.1. Dataset Description

• Dataset X1: In this study, Dataset X1, as shown in Table 6, was employed, extracted from the AHP-QFD matrix presented in Table 3. Dataset X1 encompasses customer requirements, denoted as X1CRi, alongside their corresponding feature requirements, denoted as X1FRi. These elements are pivotal in evaluating the most suitable locations for cargo logistics centers.

Table 6. Dataset X1 From AHP-QFD matrix of airport cargo site.

	X1 FR1	X1 FR2	X1 FR3	X1 FR4	X1 FR5	X1 FR6	X1 FR7	X1 FR8	X1 FR9
X1CR1	0.90	2.70	2.70	0.90	2.70	0.00	2.70	0.90	2.70
X1CR2	0.54	0.54	0.54	0.18	0.18	0.18	0.18	0.00	0.54
X1CR3	0.78	0.78	0.00	0.26	0.78	0.26	0.00	0.09	0.00
X1CR4	0.27	0.27	0.09	0.09	0.00	0.00	0.00	0.00	0.09
X1CR5	0.00	1.14	0.00	0.38	0.00	0.00	0.00	0.00	0.38
X1CR6	0.37	1.12	0.00	0.37	0.00	1.12	0.00	0.00	0.00
X1CR7	0.00	0.13	0.00	0.13	0.00	0.38	0.38	0.00	0.00
X1CR8	0.00	0.00	0.00	0.19	0.02	0.06	0.06	0.00	0.19
X1CR9	0.00	0.00	0.00	0.00	0.00	0.15	0.15	0.15	0.00
X1CR10	0.00	1.73	0.00	1.73	0.00	0.58	0.00	1.73	1.73

Note: X1FR_i is feature requirement i of Dataset X1 and X1CR_i is customer requirement i of Dataset X1.

Table 6. Dataset X1 From AHP-QFD matrix of airport cargo site.

	X1 FR1	X1 FR2	X1 FR3	X1 FR4	X1 FR5	X1 FR6	X1 FR7	X1 FR8	X1 FR9
X1CR1	0.90	2.70	2.70	0.90	2.70	0.00	2.70	0.90	2.70
X1CR2	0.54	0.54	0.54	0.18	0.18	0.18	0.18	0.00	0.54
X1CR3	0.78	0.78	0.00	0.26	0.78	0.26	0.00	0.09	0.00
X1CR4	0.27	0.27	0.09	0.09	0.00	0.00	0.00	0.00	0.09
X1CR5	0.00	1.14	0.00	0.38	0.00	0.00	0.00	0.00	0.38
X1CR6	0.37	1.12	0.00	0.37	0.00	1.12	0.00	0.00	0.00
X1CR7	0.00	0.13	0.00	0.13	0.00	0.38	0.38	0.00	0.00
X1CR8	0.00	0.00	0.00	0.19	0.02	0.06	0.06	0.00	0.19
X1CR9	0.00	0.00	0.00	0.00	0.00	0.15	0.15	0.15	0.00
X1CR10	0.00	1.73	0.00	1.73	0.00	0.58	0.00	1.73	1.73

Note: X1FR_i is feature requirement i of Dataset X1 and X1CR_i is customer requirement i of Dataset X1.

• Dataset X2: In this study, Dataset X2, as shown in Table 7 was utilized, sourced from the AHP-QFD matrix presented in Table 4. Dataset X2 consists of customer requirements, identified as X2CRi, and their corresponding feature requirements, denoted as X2FRi. The focus of Dataset X2 is on the enhancement of the current product utilized at Songklanagarind Hospital, particularly in the design of a temperature-controlled medicine bag.

Table 7. Dataset X2 From AHP-QFD controlled temperature medicine bag design.

	X2 FR1	X2 FR2	X2 FR3	X2 FR4	X2 FR5	X2 FR6	X2 FR7	X2 FR8	X2 FR9	X2 FR10	X2 FR11	X2 FR12	X2 FR13	X2 FR14
X2CR1	44.46	44.46	0.00	44.46	4.94	14.82	0.00	4.94	0.00	0.00	0.00	4.94	0.00	0.00
X2CR2	44.19	44.19	0.00	44.19	4.91	14.73	0.00	0.00	0.00	0.00	0.00	14.73	0.00	0.00
X2CR3	0.00	0.00	0.00	0.00	41.94	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
X2CR4	0.00	0.00	44.64	4.96	0.00	0.00	0.00	14.88	0.00	0.00	0.00	4.96	44.64	0.00
X2CR5	14.52	43.56	0.00	0.00	0.00	14.52	0.00	4.84	0.00	0.00	0.00	0.00	4.84	0.00
X2CR6	13.29	13.29	13.29	13.29	0.00	0.00	0.00	39.87	39.87	39.87	0.00	0.00	0.00	13.29
X2CR7	0.00	0.00	11.13	0.00	0.00	33.39	11.13	33.39	33.39	33.39	33.39	0.00	11.13	11.13
X2CR8	14.61	14.61	14.61	4.87	43.83	14.61	4.87	43.83	0.00	0.00	43.83	0.00	43.83	0.00
X2CR9	3.82	3.82	3.82	3.82	0.00	0.00	0.00	11.46	34.38	34.38	11.46	0.00	0.00	11.46
X2CR10	0.00	0.00	0.00	0.00	0.00	44.82	0.00	14.94	0.00	0.00	0.00	4.98	44.82	0.00
X2CR11	0.00	0.00	0.00	0.00	0.00	0.00	0.00	14.07	0.00	42.21	0.00	0.00	0.00	0.00
X2CR12	5.00	5.00	45.00	5.00	45.00	45.00	5.00	15.00	0.00	0.00	45.00	0.00	45.00	0.00
X2CR13	0.00	0.00	0.00	0.00	0.00	0.00	31.86	10.62	10.62	10.62	0.00	0.00	0.00	0.00
X2CR14	0.00	0.00	0.00	0.00	0.00	45.00	15.00	45.00	0.00	0.00	45.00	0.00	15.00	0.00
X2CR15	0.00	0.00	0.00	0.00	0.00	0.00	0.00	14.49	14.49	14.49	43.47	4.83	0.00	43.47
X2CR16	45.00	45.00	45.00	45.00	45.00	5.00	5.00	45.00	0.00	0.00	15.00	0.00	45.00	0.00
X2CR17	10.05	10.05	10.05	10.05	10.05	3.35	3.35	10.05	0.00	0.00	0.00	30.15	0.00	0.00

Note: X2FR_i is feature requirement i of Dataset X2 and X2CR_i is customer requirement i of Dataset X2.

Table 7. Dataset X2 From AHP-QFD controlled temperature medicine bag design.

	X2 FR1	X2 FR2	X2 FR3	X2 FR4	X2 FR5	X2 FR6	X2 FR7	X2 FR8	X2 FR9	X2 FR10	X2 FR11	X2 FR12	X2 FR13	X2 FR14
X2CR1	44.46	44.46	0.00	44.46	4.94	14.82	0.00	4.94	0.00	0.00	0.00	4.94	0.00	0.00
X2CR2	44.19	44.19	0.00	44.19	4.91	14.73	0.00	0.00	0.00	0.00	0.00	14.73	0.00	0.00
X2CR3	0.00	0.00	0.00	0.00	41.94	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
X2CR4	0.00	0.00	44.64	4.96	0.00	0.00	0.00	14.88	0.00	0.00	0.00	4.96	44.64	0.00
X2CR5	14.52	43.56	0.00	0.00	0.00	14.52	0.00	4.84	0.00	0.00	0.00	0.00	4.84	0.00
X2CR6	13.29	13.29	13.29	13.29	0.00	0.00	0.00	39.87	39.87	39.87	0.00	0.00	0.00	13.29
X2CR7	0.00	0.00	11.13	0.00	0.00	33.39	11.13	33.39	33.39	33.39	33.39	0.00	11.13	11.13
X2CR8	14.61	14.61	14.61	4.87	43.83	14.61	4.87	43.83	0.00	0.00	43.83	0.00	43.83	0.00
X2CR9	3.82	3.82	3.82	3.82	0.00	0.00	0.00	11.46	34.38	34.38	11.46	0.00	0.00	11.46
X2CR10	0.00	0.00	0.00	0.00	0.00	44.82	0.00	14.94	0.00	0.00	0.00	4.98	44.82	0.00
X2CR11	0.00	0.00	0.00	0.00	0.00	0.00	0.00	14.07	0.00	42.21	0.00	0.00	0.00	0.00
X2CR12	5.00	5.00	45.00	5.00	45.00	45.00	5.00	15.00	0.00	0.00	45.00	0.00	45.00	0.00
X2CR13	0.00	0.00	0.00	0.00	0.00	0.00	31.86	10.62	10.62	10.62	0.00	0.00	0.00	0.00
X2CR14	0.00	0.00	0.00	0.00	0.00	45.00	15.00	45.00	0.00	0.00	45.00	0.00	15.00	0.00
X2CR15	0.00	0.00	0.00	0.00	0.00	0.00	0.00	14.49	14.49	14.49	43.47	4.83	0.00	43.47
X2CR16	45.00	45.00	45.00	45.00	45.00	5.00	5.00	45.00	0.00	0.00	15.00	0.00	45.00	0.00
X2CR17	10.05	10.05	10.05	10.05	10.05	3.35	3.35	10.05	0.00	0.00	0.00	30.15	0.00	0.00

Note: X2FR_i is feature requirement i of Dataset X2 and X2CR_i is customer requirement i of Dataset X2.

• Dataset X3: In this study, Dataset X3, as shown in Table 8 was employed, derived from the AHP-QFD matrix presented in Table 5. Dataset X3 includes customer requirements, denoted as X3CRi, and their associated feature requirements, labeled as X3FRi. Notably, Dataset X3 introduces an approach to new product development and showcases its application through a case study centered on the development of new sports earphones.

Table 8. Dataset X3 From AHP-QFD new earphone development.

	X3 FR1	X3 FR2	X3 FR3	X3 FR4	X3 FR5	X3 FR6	X3 FR7	X3 FR8	X3 FR9	X3 FR10
X3CR1	0.00	0.00	0.07	0.00	0.00	0.00	0.00	0.00	0.07	0.00
X3CR2	0.00	0.06	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.17
X3CR3	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.17
X3CR4	0.05	0.05	0.00	0.08	0.00	0.00	0.00	0.00	0.00	0.00
X3CR5	0.20	0.30	0.00	0.00	0.00	0.00	0.00	0.10	0.20	0.00
X3CR6	0.00	0.11	0.00	0.00	0.00	0.00	0.00	0.07	0.07	0.00
X3CR7	0.00	0.00	0.00	0.00	0.23	0.15	0.00	0.00	0.00	0.00
X3CR8	0.48	0.48	0.00	0.00	0.00	0.00	0.73	0.00	0.00	0.00

Note: X3FR_i is feature requirement i of Dataset X3 and X3CR_i is customer requirement i of Dataset X3.

Table 8. Dataset X3 From AHP-QFD new earphone development.

	X3 FR1	X3 FR2	X3 FR3	X3 FR4	X3 FR5	X3 FR6	X3 FR7	X3 FR8	X3 FR9	X3 FR10
X3CR1	0.00	0.00	0.07	0.00	0.00	0.00	0.00	0.00	0.07	0.00
X3CR2	0.00	0.06	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.17
X3CR3	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.17
X3CR4	0.05	0.05	0.00	0.08	0.00	0.00	0.00	0.00	0.00	0.00
X3CR5	0.20	0.30	0.00	0.00	0.00	0.00	0.00	0.10	0.20	0.00
X3CR6	0.00	0.11	0.00	0.00	0.00	0.00	0.00	0.07	0.07	0.00
X3CR7	0.00	0.00	0.00	0.00	0.23	0.15	0.00	0.00	0.00	0.00
X3CR8	0.48	0.48	0.00	0.00	0.00	0.00	0.73	0.00	0.00	0.00

Note: X3FR_i is feature requirement i of Dataset X3 and X3CR_i is customer requirement i of Dataset X3.

The AHP-QFD matrix defines the input dataset, X ∈ R_m×n, for the feature selection algorithm. It should be noted that the customer requirements are the m observations of X, and the feature requirements are the n observations of X. The list of customer requirements and feature requirements associated with these AHP-QFD matrices are provided in

Table 9. The list of customer requirements.

* CRi	X1	X2	X3
CR1	Quick/accurate logistic ability	Keep Cool	Lightweight
CR2	Comfortable logistic	Cooling generator	Hold position
CR3	Adequate building space	Temperature display	Wireless
CR4	Proper logistics center site	Anti-shock	An open ear for health
CR5	Reasonable logistics fee	External heat protector	An open ear for safety
CR6	Easy to load/unload	Temperature display	Waterproof
CR7	Adequate airport service supply	Bag form	Comfort
CR8	Sufficient public facilities	Trustworthy	Easy to carry
CR9	Stable economic utilities provide	Lightweight	Continuous usage
CR10	Quick delivery service capability	Covering	Cool appearance
CR11		Sufficient capacity
CR12		Safety
CR13		Easy to carry
CR14		Easy to clean
CR15		Suitable price
CR16		Quality material
CR17		Other utilities

* CR_i refers to customer requirement i of each dataset.

and

Table 10. The list of feature requirements.

* FRi	X1	X2	X3
FR1	Airport land feature	Insulation ability of the inner layer	Material
FR2	Considering airport operating handling cost	Heat reflection ability of the outer layer	Structure
FR3	Airport carrying condition	Anti-shock ability	Wireless
FR4	Closeness to airline agent and customer	Cooling ability	Sealing
FR5	Local political regulation and law	Temperature indicator	Battery capacity
FR6	Improvement district’s working	Ziploc enclosure	Power consumption
FR7	Airport labor status	Carry handle	Appearance
FR8	Airport electrical power/utilities	Bag form	Weight
FR9	Airport information technology status	Weight	Connection wire length
FR10		Size	Diaphragm design
FR11		Medical disposable item
FR12		Reusable
FR13		Safety material
FR14		Price

* FR_i refers to feature requirement i of each dataset.

.

Table 9 provides a detailed list of customer requirements across the three datasets. Each customer requirement represents a specific need or expectation that must be addressed in the context of the respective dataset.

Table 10 lists the feature requirements corresponding to the customer requirements from the same three datasets. Each feature requirement represents a specific attribute or characteristic that a product or service must possess to satisfy the associated customer requirements.

The matrices in Table 6, Table 7 and Table 8 were appropriately formatted as input datasets, X ∈ R_m×n, for the feature selection algorithm by the PCA method. And the weights of each requirement derived from AHP were included in the dataset to enhance the processing data.

3.3.2. PCA Method

• Data Preprocessing: Before performing PCA, the datasets were standardized to ensure that each feature contributed equally to the analysis. Standardization involves scaling the features to have a mean of 0 and a standard deviation of 1. As the code below:
  • data_x1 = np.array([Table 6])
  • data x2 = np.array([Table 7])
  • data x3 = np.array([Table 8])
  • data_(x1,x2,x3)_standardized = (data_(x1,x2,x3) − np.mean(data_(x1,x2,x3) axis = 0))/np.std(data_(x1,x2,x3), axis = 0)
• PCA Implementation: PCA was implemented using the PCA class from the sklearn.decomposition module in Python. The absolute loadings of the first principal component were calculated to determine the contribution of each feature to the principal component. As the code below:
  • from sklearn.decomposition import PCA
  • pca_x1 = PCA()
  • pca_x1.fit(data_x1_standardized)
  • pca_x2 = PCA()
  • pca_x2.fit(data_x2_standardized)
  • pca_x3 = PCA()
  • pca_x3.fit(data_x1_standardized)

3.4. Evaluation of Interpretability and Usability for Decision-Making by Product Designers

To assess the interpretability and usability of the PCA-transformed matrices for decision-making by product designers, a qualitative approach was employed. Product designers were presented with both the original AHP-QFD matrices and the PCA-transformed matrices derived from the datasets of Case study A (Logistics Industry: X1-Airport Cargo Logistic Center Site) and Case study B (Consumer Electronics Industry: X3-New Earphone Development). They were tasked with evaluating the clarity, comprehensibility, ease of deployment, and usefulness of the matrices in guiding their decision-making process regarding feature prioritization in product design.

Semi-structured interviews were conducted to gather in-depth insights and feedback from the product designers. During these interviews, designers were encouraged to elaborate on their perceptions, highlighting any challenges or advantages they encountered when utilizing the PCA-transformed matrices. Additionally, designers were invited to provide feedback on the ease of deployment and understandability of the integrated PCA-AHP-QFD approach. The evaluation process involved a series of structured questions aimed at comparing Method 1 (AHP-QFD) and Method 2 (PCA integrated with AHP-QFD) across various criteria.

4. Results

4.1. Dataset X1

4.1.1. Scree Plot Analysis for Dataset X1

The scree plot generated from the PCA results in Figure 8 provides insights into the variance explained by each principal component. In the case of Dataset X1, the scree plot revealed the following:

Figure 8 illustrates the scree plot, providing insights into the variance explained by each principal component. Key observations for Dataset X1 include the following:

• Dominant First Component: The first principal component elucidates a significant portion of the dataset’s variance.
• Elbow Point: A distinct “elbow” in the scree plot suggests a clear cutoff point for dimensionality reduction.
• PCA Suitability: Given the substantial variance explained by the first component, PCA is deemed suitable for reducing the dimensionality of Dataset X1.

4.1.2. Cumulative Explained Variance Ratio for Dataset X1

Analysis of the explained variance ratio for Dataset X1 yields the following insights:

• The first principal component accounts for 60% of the dataset’s variance.
• The first two principal components collectively explain 83.4% of the variance.
• By the fifth principal component, 99.2% of the variance is explained.

4.1.3. Feature Contribution for Dataset X1

Absolute loadings of the first principal component reveal the significance of each feature. Notable feature contributions for Dataset X1 are depicted in Figure 9.

The absolute loadings of the first principal component in Dataset X1 offer valuable insights into the relative importance of each feature requirement for optimal cargo logistics center locations. The following detailed analysis highlights the significance of each feature.

Among the features examined, FR9 emerges as the most critical factor, with its highest loading indicating its pivotal role in assessing suitability. Additionally, FR2, FR3, FR5, and FR7 exhibit substantial loadings, highlighting their significant influence on decision-making. Conversely, FR4, FR8, and FR1 demonstrate moderate loadings, suggesting their relevance but lesser importance compared to the aforementioned factors. Finally, FR6, focusing on the improvement district’s work, emerges as the least significant, with the lowest loadings indicating its relatively minor impact on cargo logistics center assessments. This comprehensive analysis enables stakeholders to prioritize factors that most significantly impact decision outcomes, thereby facilitating more informed and effective decision-making processes in cargo logistics center assessments.

4.2. Dataset X2

4.2.1. Scree Plot Analysis for Dataset X2

In Figure 10, the scree plot generated from the PCA results provides insights into the variance explained by each principal component.

Figure 10 illustrates the scree plot, providing insights into the variance explained by each principal component. Key observations for Dataset X2 include:

• Gradual Decline: Variance explained by principal components exhibits a gradual descent, indicating an even distribution of information.
• No Clear Elbow: Unlike Dataset X1, Dataset X2 lacks a distinct “elbow” in the scree plot, making the optimal number of components challenging.

4.2.2. Cumulative Explained Variance Ratio for Dataset X2

The first component explains 27.3% of the variance, and the first 7 components together explain 90.6% of the variance.

4.2.3. Feature Contribution for Dataset X2

In Figure 11, the feature contributions for Dataset X2 are depicted, outlining significant features and their implications.

The absolute loadings of the first principal component in Dataset X2 illuminate the significance of each feature requirement in the design of temperature-controlled medicine bags for Songklanagarind Hospital. Through a comprehensive analysis of these feature contributions, several key insights emerge:

Firstly, FR1 emerges as the highest priority feature, with its highest loading underscoring its pivotal role in ensuring the effectiveness of temperature control in medicine storage. This emphasizes its critical importance in maintaining the integrity of medical supplies. Furthermore, FR4 and FR2 are identified as high-priority features, with substantial loading values highlighting their significance in protecting fragile medical equipment from damage during transport. Additionally, FR10 and FR9, focusing on size and weight, demonstrate notable loadings, indicating their significant impact on the design and usability of medicine bags. Factors such as portability, storage space optimization, and user convenience contribute to their importance in meeting operational needs.

On the other hand, FR14, FR5, FR13, and FR3 are deemed to have moderate importance, with moderate loadings suggesting their relevance but lesser priority compared to the critical features outlined above.

Moreover, FR12 and FR7 exhibit mid-range loadings, reflecting their moderate influence on product design. While contributing to the overall assessment, they are considered to have a lesser extent of importance.

Lastly, FR6 and FR8, focusing on Ziploc enclosure and bag form, are identified as the least priority features, with the lowest loadings indicating their relatively minor impact on medicine bag design. Although offering convenience, they are not as critical as cooling and shock-proofing elements.

This detailed feature analysis provides valuable insights for optimizing the design of temperature-controlled medicine bags, enabling stakeholders to prioritize features based on their impact on product performance and user satisfaction.

4.3. Dataset X3

4.3.1. Scree Plot Analysis for Dataset X3

In Figure 12, the scree plot generated from the PCA results provides insights into the variance explained by each principal component.

Figure 12 illustrates the scree plot for Dataset X3, offering insights into variance explained by each principal component:

• High Initial Variance: Approximately 30% of the variance is explained by the first principal component, signifying substantial capture of dataset information.
• Noticeable Decline: A discernible decline post the first component suggests diminishing contributions from subsequent components.
• PCA Appropriateness: Given the high variance explained by the first component, PCA appears appropriate for dimensionality reduction while retaining crucial information.

4.3.2. Cumulative Explained Variance Ratio for Dataset X3

The first component explains 32.9% of the variance, with the first four components collectively elucidating 86.9% of the variance.

4.3.3. Feature Contribution for Dataset X3

In Figure 13, the feature contributions for Dataset X3 are depicted, outlining significant features and their implications.

The absolute loadings of the first principal component in Dataset X3 offer valuable insights into the relative importance of each feature requirement in the development of new sports earphones. A detailed analysis of the feature contributions reveals the following insights.

Firstly, FR2 focusing on structure for support and stability, emerges as the most significant feature, with the highest loading indicating its pivotal role in product development. Moreover, FR1 (material for durability and quality) and FR7 (appearance for aesthetic appeal) are identified as highly significant features, exhibiting substantial loading values that highlight their importance in ensuring product quality and visual appeal.

Additionally, FR10 (diaphragm design), FR9 (connection wire length), FR8 (weight for comfort and wearability), FR6 (power consumption for battery efficiency), and FR5 (battery capacity for usage duration) demonstrate notable loadings, indicating their significant impact on product performance and user experience. While exhibiting moderate loading, they are considered less significant than FR2, FR1, and FR7.

On the other hand, FR3 exhibits a relatively lower loading, suggesting its lesser impact on product performance. Meanwhile, FR4 has the lowest loading, indicating its minimal impact on overall product quality in this context.

This detailed feature analysis provides valuable guidance for prioritizing feature requirements in sports earphone development, enabling stakeholders to focus on factors that most significantly impact product performance, user experience, and market competitiveness.

4.4. Evaluation of Interpretability and Usability for Decision-Making Results

4.4.1. Clarity

In evaluating the clarity of the methods, experts provided insights specific to each case study. For Case Study A in the logistics industry, the initial AHP-QFD matrix identified and prioritized key site selection attributes, including operating handling costs, information technology status, and proximity to airline agents and customers. These attributes were complex and required detailed analysis, making it difficult for decision-makers to interpret the results clearly. However, after integrating PCA, 75% of the experts found Method 2 to be clearer. The PCA process simplified the complexity by reducing redundancy and highlighting the most critical components, making the evaluation process more straightforward and easier to understand.

In Case Study B for new earphone development, the initial AHP-QFD matrix outlined numerous technical requirements for the product, such as material, structure, and wireless capabilities. The detailed technical specifications made the process intricate and harder to interpret. With PCA integration, 75% of the experts found Method 2 clearer. The PCA helped simplify these detailed specifications by focusing on the most influential factors, thus enhancing the clarity of the decision-making process.

4.4.2. Speed

The speed of evaluating feature requirements is a critical factor in decision-making efficiency. In Case Study A for the logistics industry, the initial AHP-QFD matrix required significant time to analyze multiple criteria for site selection. The detailed analysis of factors was time-consuming. After PCA integration, 100% of the experts preferred Method 2 for its ability to evaluate important functional requirements more quickly. PCA reduced the evaluation time by focusing on the most critical components, such as operating handling costs and information technology status, thus speeding up the overall evaluation process.

In Case Study B for new earphone development, the initial matrix involved detailed technical specifications for earphone design, which took considerable time to evaluate. With PCA integration, 87.5% of the experts found Method 2 quicker for evaluating important feature requirements. The PCA process expedited the identification and evaluation of key specifications, such as battery capacity and weight, making the evaluation process faster and more efficient.

4.4.3. Accuracy

Accuracy is paramount in decision-making processes, ensuring that the most critical features are prioritized correctly. In Case Study A for the logistics industry, the initial AHP-QFD matrix provided a comprehensive analysis of site selection criteria, but the complexity made it difficult to ensure precise prioritization. After integrating PCA, 75% of the experts deemed Method 2 more accurate. PCA allowed for a more precise evaluation by eliminating redundant data and highlighting the most important factors.

In Case Study B for new earphone development, the initial matrix detailed numerous technical aspects of earphone design, making accurate prioritization challenging. With PCA integration, 62.5% of the experts believed that Method 2 allowed for a more accurate evaluation of feature requirements. The PCA process improved accuracy by concentrating on the most influential factors, such as material and wireless capabilities.

4.4.4. Overall Effectiveness in Reducing Evaluation Time and Appropriateness of Methods

The experts generally agreed that PCA integrated with AHP-QFD could significantly reduce the time required for evaluating new product and service designs. Specific estimates and opinion are included in

Table 11. Expert opinions on the effectiveness and appropriateness of method.

Participant Expertise	Opinion
Environmental Technical Artists in the Game Industry with Architectural Design Knowledge (6 Years’ Experience)	Method 2 is faster because it evaluates a single dimension. However, Method 1 might provide more accurate answers in specific contexts, saving time in the long run. Method 1 is more comprehensive; early-stage analysis is crucial, and although it takes more time initially, it leads to better, more accurate development.
Senior IT Specialist/robotic system design (15 Years’ Experience)	Method 2 can reduce evaluation time by approximately 90%. Method 2 is better for evaluation and decision-making, but Method 1 emphasizes detailed requirements.
IT/Front-End Developer (8 Years’ Experience)	Method 2 can reduce evaluation time by 70%. Method 2 is faster for comparative results and finding solutions at every stage.
Product and Innovation Design Research Center (4 Years’ Experience/2 Years’ Experience)	Method 2 is more effective because its results are less complex, making it faster to evaluate key features. Within limited time, Method 2 provides more accurate evaluations. However, with unlimited time, both methods are equally accurate, though Method 1 requires more experience and time. Method 2 remains a suitable choice because its results are quicker to interpret, reducing time. However, if the decision-maker lacks experience, Method 1 might be more accurate but requires more time.
Interaction Design and Innovation Master Students (3 Years’ Experience in Banking system, 2 Years’ Experience as UX/UI design student)	Method 2 is quicker for identifying key requirements due to its visual and ordered presentation. Method 2 can reduce time by 40%, though it might require an additional 20% for re-evaluation. Method 1 is better for detailed understanding. Method 1 should still be prioritized for deep understanding, even though Method 2 helps reduce time. Method 1 is suitable for ministerial-level analysis, while Method 2 is for managerial and decision-making levels.
Innovation Designer (10 Years’ Experience)	Method 2 can reduce the time needed for defining important features by 70%. Method 2 might take slightly more time in pre-analysis but benefits the design phase by reducing overall time.
Educational Designers (33 Years’ Experience)	Method 2 can reduce time by 90% due to its straightforward and clear results, which are easier and faster to interpret, making it more efficient. Method 2 is more efficient, even if it takes slightly more time in pre-analysis, as it provides 90% efficiency in aligning with CRs.
Traffic and Route Designers (41 Years’ Experience)	Method 2 can reduce time by 95% due to its clear and quick interpretation of results. Method 2, despite taking more time in preparation, offers clearer and more efficient results, highly favored by clients.

.

Participants were chosen based on their extensive experience in their respective fields related to product and service design. They were selected from automotive part companies, logistics organizations, research and development organizations, game and multimedia companies, and educational centers.

The primary selection was conducted via a Google Form questionnaire, and those without experience in service and design objectives were excluded from the interview and evaluation questionnaire. The selected participants come from diverse backgrounds with substantial experience in their respective fields. This includes environmental technical artists in the Game Industry with architectural design knowledge, Senior IT specialist/robotic system design, IT/front-end developer, product and innovation design research center, interaction design and innovation master students, innovation designer, educational designers, and traffic and route designers. Their years of experience range from 2 to 41 years. This variety ensures a comprehensive perspective on the effectiveness of the methods.

The interviews were meticulously constructed to capture detailed opinions on the effectiveness of the two methods. Each participant was provided with a comprehensive overview of both methods and their respective processes. Structured questions were designed to elicit specific feedback on the time efficiency, accuracy, and overall effectiveness of each method. Follow-up questions were used to delve deeper into their initial responses, ensuring a thorough understanding of their perspectives.

These opinions collectively highlight the strengths of both methods, with Method 2 offering significant time savings and ease of use, while Method 1 provides the depth required for comprehensive analysis in early product development stages.

5. Discussion

5.1. Applicability of PCA-AHP-QFD Integration in Different Study Industries

This section evaluates how the integration of PCA with AHP-QFD is appropriate for the selected study industries, considering their unique challenges and the potential benefits of the methodology.

• Logistics Industry
  Key Challenges: The logistics industry involves complex, multifaceted operations requiring precise coordination and efficient resource utilization. Prioritizing features that optimize cargo handling and storage while ensuring compliance with stringent regulatory standards is a significant challenge [56]. Additionally, logistics service centers must be resilient to disruptions caused by events such as natural disasters or operational failures, which can lead to severe congestion and reduced service quality [57]. The selection of an optimal site for airport cargo logistics centers also poses a strategic challenge, requiring consideration of multiple objectives and constraints such as budget allocation, customer requirements, and operational efficiency [23]. Furthermore, the logistics companies in urban areas must adapt to regulations aimed at reducing environmental impacts, such as minimizing CO₂ emissions and managing urban traffic congestion, which add another layer of complexity to operations [58].
  Applicability of PCA-AHP-QFD: Integrating PCA with AHP-QFD in this context allows for clearer identification and prioritization of key features by reducing the complexity of the AHP-QFD matrices. Expert feedback supports this ability, with 75% of participants preferring this method for its clarity and accuracy, and 100% for speed. The integration also enhances the system’s resilience by optimizing the configuration of logistics service centers to withstand disruptions. Additionally, this method can support sustainable logistics practices by providing a structured approach to evaluating and selecting appropriate strategies for logistics operations. This kind of approach takes into account various stakeholder needs and ensures alignment with organizational objectives [59].
• Personal/Consumer Electronics Industry
  Key Challenges: The personal and consumer electronics industry faces significant challenges in designing new products, some of which have already been suggested in the literature review part. Prioritizing customer needs is crucial for developing successful products [7,24,38,60]. Effectively addressing these needs can provide a market advantage, as products aligned with customer desires are more likely to succeed [24,35,36,38]. Recent trends show increasing product requirements and rapidly evolving markets demanding precise developments in shorter cycles [8,9]. This complexity necessitates precise designs with numerous variables, making design decisions more challenging and extending development times [8,9,10,11].
  Applicability of PCA-AHP-QFD: The application of PCA simplifies the complexity of AHP-QFD matrices, aiding in the prioritization of critical features with the first four components explaining 86.9% of the variance in Dataset X3. This approach facilitates efficient decision-making in new product development, reduces development cycle time, and enhances alignment with customer needs, thereby improving market acceptance and competitiveness. According to the expert feedback, 75% found it clearer, 87.5% preferred the speed of evaluations with PCA, and 62.5% found it more accurate. To effectively meet the challenges previously mentioned, methodologies like PCA-AHP-QFD can systematically prioritize and integrate customer requirements into the design process [60,61].
• Healthcare/Medical Industry
  Key Challenges: The healthcare/medical industry demands high precision and reliability in product design to ensure patient safety and regulatory compliance. The design and development of medical devices, such as temperature-controlled medicine bags, must account for various factors including user safety, efficacy, and compliance with healthcare regulations [30]. Additionally, there are substantial challenges in incorporating patient-centric designs that address the needs of diverse patient populations, particularly older adults and those with chronic conditions [62]. Healthcare providers face increasing complexity due to the evolving nature of diseases and the need for personalized medicine. This complexity necessitates robust design methodologies that can manage multiple variables and ensure the highest quality standards [63]. The importance of integrating feedback from patients and healthcare professionals into the design process to enhance the usability and effectiveness of medical devices is also highlighted [62]. Furthermore, the development of intelligent healthcare products, such as home blood glucose monitors, highlights the need for aging-friendly design that aligns technological advancements with user-centric principles, ensuring accessibility and usability for elderly users [64].
  Potential of PCA-AHP-QFD: Integrating PCA with AHP-QFD in the healthcare sector can significantly streamline the feature prioritization process by reducing the complexity of the AHP-QFD matrices. Although Dataset X2 showed a more even distribution of variance across principal components, indicating a less dominant first component, PCA still aids in dimensionality reduction. This is crucial for managing the complexity of medical device design. By simplifying the dimensionality, PCA facilitates the identification of key features that meet both patient and regulatory requirements. However, the gradual decline in explained variance suggests that PCA may need to be supplemented with additional analytical methods to fully address the specific needs of this sector.

5.2. The Fitness of PCA Model Intregrate with AHP-QFD Datasets across Study Industries

PCA consistently demonstrated effectiveness in reducing the complexity of AHP-QFD matrices, yet variations were observed in its suitability across different industrial sectors and datasets. For instance:

• Logistics Industry (Case Study A: Airport Cargo Logistic Center Site, Dataset X1): The scree plot for Dataset X1 revealed a dominant first component that explained a significant portion of the dataset’s variance. The first two principal components accounted for 83.4% of the variance, and the first five components collectively explained 99.2% of the variance. This indicates that PCA is highly suitable for dimensionality reduction in this dataset.
• Personal/Consumer Electronics Industry (Case Study B: New Earphone Development, Dataset X3): The scree plot for Dataset X3 showed that approximately 30% of the variance was explained by the first principal component, with a noticeable decline in variance explained by subsequent components. The first four components collectively elucidated 86.9% of the variance, making PCA appropriate for this dataset.
• Healthcare/Medical Industry (Dataset X2: Temperature-Controlled Medicine Bags): The scree plot for Dataset X2 exhibited a gradual decline in variance explained by principal components, with no clear “elbow” point. The first component explained 27.3% of the variance, and the first seven components together explained 90.6% of the variance. This even distribution of information suggests that PCA might not be as beneficial for this dataset. This observation is consistent with the findings of construction automation industry, which noted that PCA might not be suitable for datasets with less dominant principal components [48]. This underscores the importance of considering dataset characteristics when applying PCA.

5.3. Expert Opinions on PCA Integration with AHP-QFD for Product Design Decision-Manking

The structured interviews with experts from various fields provided valuable insights into the applicability and limitations of PCA integrated with AHP-QFD. Environmental technical artists in the game industry highlighted that while the PCA-AHP-QFD method (Method 2) is faster, the traditional AHP-QFD method (Method 1) might provide more accurate answers in specific contexts. Senior IT specialists and front-end developers noted significant time reductions (up to 90%) with Method 2, emphasizing its effectiveness for quick evaluations.

Experts from the Product and Innovation Design Research Center mentioned that Method 2 is more effective for quick evaluations, but Method 1 might be better for detailed analysis when more time is available. Interaction design and innovation master students indicated that Method 2 is suitable for managerial decision-making due to its speed and clarity.

Converging opinions from environmental technical artists, senior IT specialists, front-end developers, and traffic and route designers generally found PCA-AHP-QFD to be beneficial in improving evaluation speed and clarity. They appreciated the method’s ability to streamline decision-making processes, especially in the early design phases.

However, diverging opinions from product and innovation design researchers, interaction design and innovation students, and educational designers acknowledged the effectiveness of PCA-AHP-QFD in reducing complexity and improving prioritization. They pointed out potential limitations in capturing detailed requirements. For phases requiring intricate detail, such as prototype development or testing, these experts suggested integrating complementary methods alongside PCA-AHP-QFD to ensure comprehensive decision-making.

6. Conclusions

The integration of PCA with AHP-QFD offers a robust approach to improving decision-making processes and feature prioritization across different industries. This study evaluated the applicability of PCA-AHP-QFD integration in the logistics and personal/consumer electronics industries, considering their unique challenges and potential benefits.

In the logistics industry, PCA-AHP-QFD proved highly effective in simplifying complex AHP-QFD matrices and enhancing system resilience by optimizing logistics service centers. Expert feedback indicated a preference for this method due to its clarity, speed, and accuracy, supporting sustainable logistics practices and streamlined decision-making.

For the personal/consumer electronics industry, the integration of PCA with AHP-QFD facilitated efficient decision-making, reduced development cycle time, and improved alignment with customer needs. Expert feedback highlighted significant improvements in evaluation speed and clarity, making this methodology suitable for rapidly evolving markets with high product demands.

In the healthcare/medical industry, the PCA-AHP-QFD model showed limitations. While it helped streamline feature prioritization and manage design complexity, the distribution of variance in the datasets indicated that PCA might not be as beneficial in this sector. This underscores the importance of considering dataset characteristics when applying PCA.

Expert feedback from various fields provided valuable insights into the practical application of PCA-AHP-QFD across different industries. There were both convergences and divergences in opinions. These diverse perspectives underscore the importance of tailoring the PCA-AHP-QFD methodology to specific industry needs and project phases, ensuring comprehensive and effective decision-making throughout the product development lifecycle.

Expert feedback was gathered for the logistics and personal/consumer electronics industries, revealing insights into the strengths and limitations of the PCA-AHP-QFD approach. Experts from various fields generally found PCA-AHP-QFD beneficial in improving evaluation speed and clarity, particularly in the early design phases. However, for phases requiring intricate detail, such as prototype development or testing, complementary methods were suggested to ensure comprehensive decision-making.

In summary, the PCA-AHP-QFD methodology demonstrates significant potential in enhancing feature prioritization and decision-making efficiency in the logistics and personal/consumer electronics industries. However, the healthcare/medical industry may require supplementary methods to address its specific challenges. Future research should focus on integrating complementary methods to further enhance the applicability of PCA-AHP-QFD in complex product design scenarios across different sectors.

The practical implications for industry practitioners are profound. Companies can leverage PCA-AHP-QFD to streamline their product design processes, particularly where feature importance is concentrated and rapid decision-making is crucial. Key steps for implementation include understanding the methodology, training the team, collecting and preparing high-quality data, applying PCA to identify key variables, integrating with AHP-QFD for prioritization, and continuously monitoring and iterating the design based on feedback. This integrated approach not only enhances feature selection efficiency and alignment with customer needs but also drives innovation and competitiveness in the market.

7. Limitations and Future Research Directions

The findings suggest that PCA may not be universally applicable across all phases of product design, particularly in industries where required characteristics are less dominant among feature characteristics such as the healthcare/medical industry. For the limitations observed with Dataset X2, PCA may not be suitable for phases that require a nuanced understanding of diverse feature requirements. In this context, product design teams should consider using other techniques alongside PCA to capture the detailed needs of the product design process.

Additionally, assumptions made during PCA application and potential biases in the data used could impact the results. Ensuring data quality and considering alternative methods can help mitigate these limitations.

Future research could explore the underlying factors contributing to the varying compatibility of PCA with different industrial sectors and phases of product design. Detailed studies focusing on specific industries could provide deeper insights into the impact of this methodology. Additionally, research should examine the impact of PCA-AHP-QFD on different dimensions such as cost and time-to-market to develop a more comprehensive understanding of its benefits and limitations.