A state-of-art review on supplier selection problem

Article history: Received October 2, 2012 Received in Revised Format January 23, 2013 Accepted February 15, 2013 Available online February 23 2013 Many supplier selection problems are involved with various criteria such as quality of supplier, price, delivery time, etc. This paper presents a survey on the implementation of using different multi criteria decision making (MCDM) methods for supplier selection problems. The reviews covers recent advances of MCDM techniques such as data envelopment analysis (DEA), analytical hierarchy process, etc. over the period 2000-2012. The review also reveals that nearly 60% of the applications are associated with business unit, 15% is related to economy, 9% is devoted to service and development and 8% is dedicated to research and development. In our survey, DEA has become the most popular technique for supplier selection problem followed by Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and Analytical Hierarchy Process (AHP).


Introduction
Supplier selection is one of the most important issues in any production units and choosing a good supplier may help increase productivity and efficiency of business units.During the past few years, there have been different methods associated with supplier selection problems.Supplier selection plays an important role on efficiency of production planning.There are many methods for supplier selection and most of them are involved using multi criteria decision making.DEA is one of the most important techniques for measuring relative efficiencies of various suppliers.In this paper, we review recent advances of different techniques of implementing different techniques such as DEA methods for supplier selection problem.During the past few years, there have been tremendous efforts on using DEA to measure relative efficiencies of suppliers.Problem-solving techniques such as linear programming, test assumptions, and methods for optimizing multiple criteria decision making (MCDM) are used to obtain good solutions.MCDM is the argument that the decision-making (DM) process in the presence of different criteria sometimes incompatible with each other methods.One problem with this method is the selection of an appropriate procedure in hypothetical situations.This section attempts to briefly describe the various methods of problem solving.MCDM models can be divided into two categories as follows,

i) Multi objective decision making (MODM)
Multi objective decision making (MODM) method where we consider different objective functions such and plan to find so called Pareto solutions.There are different approaches to handle MODM problems such as goal programming, Lexicography, LP-norm, etc.

ii) Multi attribute decision making (MADM)
In multi attribute decision making (MADM), we look to rank different alternatives based on various attributes.There are literally different methods for ranking suppliers based on multiple criteria such as quality, price, etc.

Lexicography
The primary objective of Lexicography is to prioritize different objectives based on their relative importance.Korhonen and Siitari (2007) used lexicographic parametric programming to rank efficient units in the DEA model.They received the efficiency curve, which was traversing through the efficient frontier from unit to other unit and used the parameterization of the right-hand side vector of the cover problem.Lozano and Villa (2009) implemented AHP and lexicographic to specify a priori a set of preference levels.

Analytical Hierarchy Process (AHP)
Analytical hierarchy process (AHP) is another multi-criteria decision making problem where we ask decision maker to give his/her judgment about the relative priority of one alternative versus another one.There are literally many applications of AHP implementation for supplier selection problem.Sevkli et al. (2007) stated that DEA hierarchy process (DEAHP) method had better performance compared with AHP method for supplier selection.In other words, DEAHP model was more cumbersome to apply and was more suitable for applications of high-value components where careful purchasing criteria were required.They suggested that AHP model used for relatively lower value components.Kuo et al. (2010) used a hybrid of AHP and DEA for developing performance evaluation to make the supplier selection decision.They used the Fuzzy AHP (FAHP) method to find the indicators' weights through expert questionnaire survey and then, these weights were integrated with fuzzy DEA.Then, they used of α-cut set and extension principle of fuzzy set theory to simplify the fuzzy DEA.Finally, fuzzy ranking using maximizing and minimizing set method was used to rank the evaluation samples.Zeydan et al. (2011) proposed new method for selecting supplier selection and evaluation quality.They used FAHP to fine criteria weights and then fuzzy TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) to utilize in finding the ranking of suppliers.So, qualitative variables were transformed into a quantitative variable for using in DEA methodology as an output called quality management system audit.Zhang et al. (2012) used combine the DEAHP model and activity-based costing (ABC) for supplier evaluation.This model, make it easy for making decisions on supplier selection and order quantity could easily be made within an integrated single objective function, which is based on consideration of the budget of the buyer and of the capacity of the supplier.This model was based on consideration of the budget of the buyer and of the capacity of the supplier.Kang et al. (2010) applied DEA to evaluate quantitative factors, and the results were transformed into pairwise comparison values for FAHP analysis.Qualitative factors evaluated through FAHP analysis.Lin et al. (2011) integrated DEA and AHP to evaluate the economic development achieved by local governments in China.They also used a time-scale comparison of the economic performances of local governments in China based on the malmquist productivity index (MPI), which stated that there was a trend of economic growth but the results showed that after discounting the advantages of location and political connections, the east district provinces of China did not have superior economic performance or a better MPI index, as compared with other districts.Çelen and Yalçin (2012) performed an investigation on performance assessment of Turkish electricity distribution utilities based on an application of combined FAHP/TOPSIS/DEA methodology to incorporate quality of service.

Simple Additive Weighting (SAW)
In Simple Additive Weighting (SAW) technique, the DM gives weight to each attribute and then, for each option, all values of the parameters that characterize the weight multiply the results together.
Alternative has the most points is selected as the best choice.Hosseinian et al. (2009) proposed a new method based on DEA for the weight vector derivation from the pair-wise comparison matrices in the group AHP called DEA-WDGD.The purpose of DEA-WDGD is either maximizing outputs or minimizing inputs.They also compared the DEA-WDGD with the DEA method recently proposed for weights derivation in the group AHP.The results indicated that DEA-WDGD provides better weights.The SAW method was also utilized to sum local weights without needing to normalize them.

Technique for Order Preference by Similarity to Ideal Solution (TOPSIS)
Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) chooses the alternative with the shortest distance from the positive ideal alternative and the greatest distance from the negative ideal alternative.Among those who used the DEA method TOPSIS this research can be referred to Tao et al. (2012) who proposed decision model in three parts.First, DEA was used to supply for the best composition on the performance parameters of original data; Second application of Axiomatic Fuzzy Set (AFS) theory and AHP method, the weight of each attribute was calculated and, third, TOPSIS model was applied to provide the ranking order from the best combinations based on the weights of attributes.This model made decision results more reasonable from DMs. Arabzad et al. (2013) proposed the model for choosing supplier based on Kraljic and DEA model.In this method, the purchase items classified to leverage, strategic and routine items.The purpose of this classification was coordination among the purchase items to allocate orders and adopt appropriate and suitable policies with the suppliers of related items.On the other hand, suppliers based upon their performance are categorized by TOPSIS technique.Finally, obtained ranking will be the criteria for the allocation of the purchase orders.Hemmati et al. (2013) used these two MCDM techniques (DEA and TOPSIS) to measure the relative efficiencies banks in terms of electronic payment.Three inputs with DEA methods were considered including the number of issued cards, the number of ATM machines and the number of POSs and two outputs including the number of successful ATMs and the number of successful POSs transactions.

Elimination and choice translation reality (ELECTRE)
Elimination and choice translation reality (ELECTRE) is another MCDM technique for ranking different alternatives (Bernard, 1968).According to Bı̇rgün and Cı̇han (2010) supplier selection is one of the most important issues of the purchasing departments and it is often considered as an MCDM problem.They used ELECTRE for supplier selection problem and discussed how to rank different suppliers based on this method.

Green supply chain
Decision support system (DSS) has also been used for analyzing data and making appropriate decisions for supplier selection problem.Akçay et al. (2012), for instance, developed a general DSS model to investigate the results of implementation of DEA models.They examined the proposed model for a real world project for benchmarking the vendors of a leading Turkish automotive company.
These days, green supply chain management has become a popular research area.Amindoust et al. (2013) applied DEA for supplier selection by considering environmental issues using linguistic terms based on decision makers' opinion.Kuo et al. (2010) developed a green supplier selection framework, which integrated artificial neural network (ANN) and a combination of DEA and analytic network process (ANP).The hybrid ANN-MADA method considered both practicality in traditional supplier selection criteria and environmental regulations.They reported that ANN-MADA had better power of discrimination and noise-insensitivity in evaluating green suppliers' performances.Wen and Chi (2010) developed a criteria set including green, traditional, and partnership issues for the green supplier selection problem.They used a combination of DEA, ANP as well as AHP methods to rank different suppliers.

Neural Network System
Neural networks systems (NNS) and computational methods for learning new knowledge obtained acts to more complex systems expected output responses.NNS is performance to display approximation and estimate.Çelebi and Bayraktar (2008) proposed new method of integration of neural networks (NN) and DEA for evaluating of suppliers under incomplete information of evaluation criteria.Wu (2009)

Genetic Algorithm (GA)
Genetic Algorithm (GA) is Darwin's principle of natural selection to find the optimal formula for predicting or use pattern matching.Forecasting techniques based on genetic algorithms for regressions are often a good option.Chuang at el. ( 2009) used multi-objective genetic algorithm (MOGA) to minimized time and costs of new products developed in conjunction with maximized product reliability.Lin at el. ( 2013) used a combination of the genetic algorithm (GA) and DEA to evaluate the simulation results and guide the search process and the model was applied to determine optimal resource levels in surgical services.

Conclusion
In this paper, efforts have been made for the collection of articles associated with the DEA and other problem-solving techniques in various industries.The survey has indicated that DEA method has become the most important technique for ranking supplier selection followed by TOPSIS, AHP.In addition, there were literally various techniques for ranking suppliers using a hybrid of two or more models.We anticipate to have a tremendous breakthrough on using more sophisticated approaches for supplier selection.

Fig
Fig. 1 show various appl to service a has become Order Prefer assessed supplier performance by a combine model with DEA, decision trees (DT) and NNs.This model had two steps: first, DEA model was applied and classified suppliers into efficient and inefficient clusters based on the resulting efficiency scores.Second, the model utilized firm performance-related data to train DT, NNs model and apply the trained decision tree model to new suppliers.Shi at el. (2010) proposed a model based on artificial intelligence (BP neural network) and C2R-DEA for selecting appropriate logistics suppliers.In this model they introduced details in the C2R-DEA cross-evaluation method.Samoilenko and Osei-Bryson (2010) proposed a five-step methodology based on DEA, Cluster Analysis (CA) and Neural Networks (NN).