Integration Dematel and ANP for The Supplier Selection in The Textile Industry: A Case Study

Currently, companies are required to improve supply chain performance. One of the main problems in the supply chain is the proper supplier selection. Supplier selection has an essential role in improving supply chain management performance. Supplier selection requires the proper criteria. However, the relationship between criteria is rarely considered in the selection of suppliers in the textile industry. This study tries to propose integrating the Decision Making Trial and Evaluation Laboratory (DEMATEL) and the Analytic Network Process (ANP) for supplier selection in the textile industry. Both methods are multi-criteria decision making (MCDM) tools DEMATEL is used to assess the relationship between criteria. Furthermore, ANP is used to evaluate and weigh the importance of criteria and suppliers. A case study was carried out in a textile company located in Indonesia. The results show that this procedure can identify the relationship and effect of each criterion. The results show that the product price criteria are the criteria that have the most significant weight. The criteria for conformity to specifications and consistency of quality are in second and third place. Finally, suppliers are selected based on weight assessment on each criterion by ANP.


I. INTRODUCTION 1
In the past decade, supply chain management has received much attention from the world, both researchers and industry players (Ibrahim, Putri, & Utama, 2020). The primary purpose of supply chain management is to reduce risk, improve company performance, and obtain maximum benefits (Boran, Genç, Kurt, & Akay, 2009). To achieve those, suppliers contribute significantly to supply chain management. The right supplier's selection can reduce purchase costs, increase competitiveness, and reduce risk (Alyanak & Armaneri, 2009) (Utama, 2021a). Improper supplier selection causes significant losses to the company, such as shortages of raw material inventory and delays in the production process   (Utama, 2021b). Therefore, choosing the right supplier is expected to minimize the risks that occur in the company (Limansantoso, 2013) (Utama, Baroto, Ibrahim, & Widodo, 2021).
In recent years, there has been increasing interest in supplier selection issues. One of them is in the textile industry. Several researchers have researched supplier selection in the textile and garment industry. Each study uses different criteria and methods. Several previous studies include Koprulu and Albayrakoglu (2007), which used the Analytical Hierarchy Process (AHP) method with economic aspects (quality, cost, delivery, flexibility, innovation) and social (trust). GÜNGÖR, COŞKUN, DURUR, and GÖREN (2010) used the AHP by considering economic aspects. Triple bottom elements (economic, social, and environmental) are also observed by Chan and Chan (2010). They proposed the AHP method for supplier selection. Y.-J. Chen (2011) used the Technique For Order Preference By Similarity Ideal Solution (Topsis) method. In addition, the adaptive neuro-fuzzy inference systems (Anfis) method was proposed by GüNeri, Ertay, and YüCel (2011) for supplier selection by considering economic and social aspects.
Several other procedures have proposed by the researchers include the gray system approach (Baskaran et al., 2012), Topsis and Fuzzy AHP (Kar et al., 2014), fuzzy Topsis (Jia et al., 2015) (Kargi, 2016) (Yildiz, 2016) (Fallahpour et al., 2017). The researcher also offered several integrated procedures. Guarnieri and Trojan (2019) integrated AHP and Electre-Tri by considering economic and social aspects. In addition, Çalı and Balaman (2019) developed the integration of Electre and Vikor for supplier selection. Based on several previous studies, the AHP procedures are popular procedures used by researchers to weight criteria and supplier selection in the textile industry. Unfortunately, this method does not consider the relationship between criteria in supplier selection. In the supplier selection on the textile industry, to the best of our knowledge, only Li, Diabat, and Lu (2019) study used the Decision Making Trial and Evaluation Laboratory (Dematel) method to identify criteria relation. However, this study only considers the economic aspects of supplier selection.
Based on previous research, very few approaches that consider the effects of criteria are investigated in the supplier selection problems. Even to the best of our knowledge, we found only one research study using dematel in the textile industry (Li et al., 2019). Unfortunately, that study only focussed on identifying the relationship between supplier selection criteria. The identification of the best supplier was not discussed in the study. Therefore, an approach is needed to solve the problem by considering the relationship between the criteria and the best supplier in the textile industry. This study proposes the Dematel and Analytic Network Process (ANP) method to solve supplier selection in the textile industry. Dematel is used to determine the effect between criteria. ANP is an effective weighting procedure that considers relations (Yang, Shieh, Leu, & Tzeng, 2008). Using the Dematel-ANP method has been proven effective for solving MCDM problems. This method has been successfully applied to various problems such as energy resources selection (Büyüközkan & Güleryüz, 2016), Risk Assessment (Dehdasht et al., 2017), risk management (Fazli, Mavi, & Vosooghidizaji, 2015), and evaluate the critical success factors (Nilashi et al., 2015) This study aims to integrate the Decision Making Trial and Evaluation Laboratory (DEMATEL) and the Analytic Network Process (ANP) for supplier selection in the textile industry. In this study, researchers proposed the Dematel method to measure the effect of the relationship between criteria. ANP is proposed to measure the importance of criteria based on the relationship between the criteria generated by Dematel. Compared to other techniques, this method has the advantage that Dematel provides a systematic approach to identify criteria, the relationship between criteria. Furthermore, ANP allows weighting based on interdependent relationships between criteria for decision-making. This study used triple bottom aspects such as economic, social, and environmental aspects for identifying criteria.

II. RESEARCH METHOD
This section presents a procedural framework for the proposed method in selecting textile industry suppliers and data collection for case studies.

Proposed Method Framework
This section proposes a procedure for selecting suppliers in the textile industry. This study integrates Dematel and ANP as a proposed method for solving problems. The Dematel method was first proposed by Fontela and Gabus (1974) based on the influence between criteria. In this study, Dematel was used to assess the effect of a criterion on the other criterion. Furthermore, the weighting of the importance level of the ANP criteria is based on the relationship between the criteria in the Dematel procedure. The ANP method is coined by Saaty (Saaty, 2004) based on a reciprocal relationship between criteria. The following is a supplier selection framework with Dematel-ANP integration, which can be seen in Figure 1. This procedure is modified from the framework proposed by Yang et al. (2008). Based on the supplier selection framework (Figure 1), Stage (1) is to identify the criteria. The determination of the criteria is carried out through the discussion by several experts from the company. Stage (2) is the making of a direct relation matrix. Direct relation assessment is carried out using a scale of 0,1,2,3, and 4. 0 represents not influencing, one is less influencing, two moderately affecting, three strongly influencing, and four are very influential. The assessment results of the relationship between the criteria are made into a direct relation matrix (matrix B) as presented in equation (1), where bij assesses the relationship between criterion i on criterion j.
Stage (3) is to calculate the Tresh Hold Value and create an α-cut total influence matrix. Based on the direct relation matrix, the following procedure normalizes the direct relation matrix using equations (2) and (3). Where X is the direct relation normalization matrix. The following procedure calculates the total relation matrix (T) based on equation (4). The threshold value is based on the average value of the normalization relation matrix. After the threshold value is found, the next step compares the value with the total relation matrix's value. Suppose the total relation matrix value is below the threshold value. In that case, it is considered that the effect of the criterion is low or negligible. It is considered to have a value of 0. The matrix formed from the comparison is called the α-cut total influence matrix. It is based on research by Yang et al. (2008).
Stage (4) is to create a Dematel network formed from an α-cut total influence matrix. Stage (5) performs a pairwise comparison based on the Dematel network. The pairwise comparison assessment uses a scale of 1, which shows that both criteria are equally important; 3 describes the criteria of Medium importance; 5 explains the criteria for Strong importance; 7 indicates Extreme importance; 9 describes the criteria of Extreme importance; and the values 2, 4, 6, 8 indicate the Intermediate scale. The pairwise comparison results are presented in a matrix as in equation (5), where j is the result of assessing pairwise comparison between criterion i on criterion j.
Stage (6) is calculating the weighted supermatrix and limit matrix. The equation used to perform calculations can be seen in equations (6) and (9). Where is the unweight supermatrix from the pairwise comparison.
is the sum on row i unweight supermatrix. Ts is the normalized unweighted supermatrix.
is a weighted supermatrix.

Rank suppliers
Determine the criteria weights Calculate unweighted supermatrix, weighted supermatrix and limit matrix Step (7) determines the criteria weights obtained from the normalization limit matrix using equation (10). Finally, step (8) ranks suppliers obtained from the normalization limit matrix.

B =
(1) Case Studies and Data Collection A case study was conducted in the textile industry in Cianjur, West Java, Indonesia. The data collected is based on Focus Group Discussion (FGD). FGD is a process of gathering information on a problem that is specifically researched. In the FGD, discussions were held with parties involved in selecting suppliers of raw materials for companies. These parties can assess the relationship between criteria and suppliers. 4 Respondents were selected to determine the Criteria, determine relationship criteria and evaluate the pairwise comparison matrix. The four respondents are the general manager, head of purchasing, head of the production, and logistics head.
Based on the FGD, the identification of supplier selection criteria in the textile industry is presented in Table 1. Measurement of the relationship between criteria was carried out using the DEMATEL questionnaire. The results of   Table 2.
The results of the direct relations from Dematel are made a network that is presented in Figure 2. From the dematel network, it is then used in ANP to separate suppliers. The ANP network for supplier selection problems in the textile industry is presented in Figure 3. In selecting suppliers, there are three pairwise comparison clusters used. Cluster 1 is used to assess the importance of criteria. Cluster 2 is used to determine the importance between criteria based on the relationship between the criteria produced by the Dematel procedure. Furthermore, the last (cluster 3) pairwise comparison to the supplier selection criteria for each alternative supplier. In this study, researchers used super decision software for weighting solutions using ANP.

Criteria relation Analysis
The α-cut total influence matrix value is shown in table 3. The average value of the total relationship matrix for the criteria is α = 0.157. Each value in the total relationship matrix is compared to the threshold value. These results indicate that the company capability (KP) has the highest number of relationships than other criteria followed by Price (HP), Compliance of material with specifications (KS), Ability to provide consistent quality. The results of this study contradict Li et al. (2019), who found quality criteria had the most significant influence. In this research, Company capability (KP) can determine other criteria. If the capability is low, then the consistency of quality, the accuracy of the number of orders, punctuality also decrease. The company capability (KP) criterion has the most number of relationships.
Furthermore, the permittance criteria (PI) have many implications for other criteria. Company suppliers are suppliers that come from abroad. Therefore, licensing has many links with other criteria. In flexibility (FL), this criterion assesses the supplier's ability to meet the demand for changes in quantity and time. These, of course, greatly influence other criteria. When the flexibility (FL) of the supplier is good, then other criteria such as accuracy of quantity and time are also good.
The criteria for On-time delivery (KW) are influenced by Delivery Cost (BK) and the type of transportation mode used (MT). The product price Figure 3. ANP Network Diagram criterion (HP) is included in the receiver group because the product's price is affected by the Compliance of material with specifications (KS) and services provided. When the product quality is good, the product price is high. Furthermore, if supplier licensing is well documented, then the product's cost will increase.

Influence between criteria
The assessment of the effect between criteria used the ANP method. The results of the impact between criteria are presented in the Unweighted Supermatrix table in table 4. For example, the criteria for Eco-friendly material (BR) affect product prices (HP) and Compliance of material with specifications (KS). Eco-friendly materials (BR) have a higher price than common raw materials. The Delivery Cost criterion (BK) influences On-time delivery (KW) and mode of transportation (MT). Shipping costs affect the type of transport used. These also affect the speed of delivery. The higher the costs incurred in delivery, the better the mode of transportation used. So, the product comes on time.
The payment method (CB) criteria affect the product price (HP) because payments from abroad to Indonesia are taxed. Therefore, the process of payment involves the cost of the product. Environmental-related certificate (EC) criteria affect the criteria for Eco-friendly material (BR). Suppliers who already have EC show that the materials used are safe for the environment. In the criteria for flexibility (FL), this criterion has the most significant influence on the criteria for accuracy of the order quantity (KJ). It is followed by the criteria for On-time delivery (KW). The criteria for flexibility assess suppliers in terms of suppliers' ability to meet changes in quantity and time requests. Product price criteria (HP) have the most significant influence on quality consistency criteria (KK). High product prices are directly proportional to product quality. In the company capability criteria (KP), this criterion significantly influences the specification conformity criteria. In addition, the criteria for Compliance of material with specifications (KS) have the most significant impact on the product price criteria.
The criterion for the accuracy of the number of orders (KJ) has the most significant effect on the Replacement of items product (PB). This criterion assesses the supplier's response to damaged materials and shortages. So, if the number of orders does not match, it will affect the change of goods. The on-time delivery (KW) criterion has the most significant effect on Delivery Cost (BK). The faster the delivery time, the greater the shipping cost. In the criteria of ability to provide consistent quality (KK), the product price (HP) is the most significant effect. It is because the better the quality of raw material, the higher the product price.
Some of the effects of other criteria are as follows: Criteria for the mode of transportation (MT) affect the accuracy of delivery costs (BK) and on-time delivery (KW). Performance history (PH) criteria have an impact on product prices (HP). The criterion for Replacement of items product (PB) affects The accuracy of the order quantity (KJ) and the ability to provide consistent quality (KK). Furthermore, the permittance (PI) impact Eco-friendly material (BR). It is because good and environmentally friendly raw materials indicate good licensing from the supplier.

Criteria Weight
Based on the ANP method, the supermatrix limit is obtained based on the results of the unweight supermatrix. The results of the limit matrix calculated based on the super decision software can be seen in table 5. Furthermore, the limit matrix must be normalized to become a global weight. The global weight and rank of the supplier selection criteria are shown in table 6.
These results show that the product price criterion (HP) has the most significant weight with a global weight of 0.1385 or 13.85%. The criteria of ability follow them to provide consistent quality (KK) and Compliance of material with specifications (KS) with global weights of 11.35% and 11.08%. This result is different from previous research conducted by Koprulu and Albayrakoglu (2007), Utama (2021a), and Utama et al. (2021). Their research shows that the criterion that has the highest weight is Quality. This is due to the different problems and conditions of the company. If the analysis is further, the product price criteria have many relationships with other criteria. In addition, product price (HP) has the highest weight because the company wants to minimize operations.
The criterion with the lowest weight is the Payment Method (CP) criterion. The payment method affects other criteria a little. In addition, this criterion is not influenced by other criteria.
The company continues to want to improve its performance by providing various payment methods. Therefore, the method of payment is   not a crucial problem in the company. In addition to the payment method criteria, the low weight criterion is the environmental certificate (EC). This criterion has little relationship with other criteria.
Even though the weight of this criterion is low, EC should still be an aspect that needs to be considered by companies.

Rank Supplier
From the supplier weight ranking (Table 7), it is found that supplier Y has the highest weight, namely 0.3885 or 38.85%. Supplier Y is the first ranked supplier. The second alternative is supplier Z, with a weight of 33.64%, followed by supplier X with a weight of 27.93%. This makes perfect sense based on pairwise comparisons because supplier Y has a better level of importance than other suppliers in each of the criteria.

IV. CONCLUSION
This research discussed supplier selection in the textile industry using Dematel and ANP. There are 15 criteria used in this study. Based on Dematel, the Company capabilities (KP) have the most significant influence than other criteria. Based on the ANP, the criteria that have the highest weight are product price (HP) followed by Compliance of material with specifications (KS) and Ability to provide consistent quality (KK). The supplier ranking results show that Supplier Y is in the first order, followed by supplier Z and supplier X. Some limitations in this study are definite information characteristics. In further research, it is necessary to consider the characteristics of uncertain information.