Multi-attribute and Multi-criteria Decision Making Model for technology selection using fuzzy logic

The problem of identification and elimination of useful technology is a multi-attribute, multi-valued problem which involves both tangible and intangible factors. To select the best useful technology that achieves most of the company requirements, it is essential to use an appropriate selection approach that takes into consideration the different quantitative and qualitative parameters of industry objectives and technology benefitsIn this paper, FMCDM (Fuzzy-based MultAttribute and Multi-criteria descision Method ) is described and explained . Keywords:Decision making, MADM, MCDM, aggregation. I. INTRODUCTION Decision making for selecting a technology requires a fair amount of information regarding the field in which the technology is to be selected for. For example, decision making framework for industrial managers to select a manufacturing technology that satisfies their supply chain objectives. Technology is defined as “the practical application of science to commerce or industry”. Hence, identifying the factors which influence the preference of the need of a specific technology, defining the effects of these factors on each other, and assessing the importance of them from the users’ point of view necessitate an effective decision making process. Decision making problem is described as the process of searching or finding the course of actions from a given set of feasible alternatives which maximizes or satisfies certain criteria associated to the goals intended to be achieved. The user needs to be assured that selecting the chosen technology contributes to the user’s success rather than hinder it. One of the toughest things about technology selection is that for every problem, there are several solutions and a mass of information available about them. Decision making is concerned mainly with the question which alternative or course of action should be undertaken under a specific situation by considering many aspects, including the degree of importance of each criterion. MultiAttribute Decision Making (MADM) refers to making decisions in the discrete decision spaces and focuses on how to select or to rank different pre-determined alternatives. Accordingly a MADM problem can be associated to as a problem of choice or ranking of the existing alternatives. However, the process of selecting the technology brings an attitude of uncertainty and subjectivity inherent in the human behavior due to which, fuzzy sets theory is applied in the decision making area known as Fuzzy Multi Criteria Decision Making (Fuzzy-MCDM). The main feature of using this approach for technology selection is that the impression inherent in the qualitative information regarding various technologies can be formalized by applying fuzzy sets theory. This will provide a consistent representation of qualitatively or linguistically formulated knowledge in a way that allows the use of precise operators and algorithms. It also enables to represent and to process adequately the vagueness or imprecision into the formal decision model in such a way without make a simplification, but still in an intellectually and scientifically acceptable manner. II. CHALLENGES IN NEW TECHNOLOGY SELECTION Specific problems have been encountered with the selection of new technology for development, be it technology new to the world or merely new to the industry. Here the aviation subsidiary company may not have the necessary experience, practical ‘know how’, the correct development life cycles, resources or the application techniques in place to develop the technology. Thus it has proved extremely difficult to answer key questions such as: ‘How many development loops are needed before an acceptable solution is reached?’ The answers to such questions could be more easily found if a methodology was followed during the selection of new technologies that accounted for all of the above issues and more. Such a procedure could also have an additional benefit that even if a project fails, every step of the selection process would have been recorded and so the reasons behind certain key decisions could be evaluated and the lessons learnt from any mistakes. In the aviation company, any new product within the global market, where performance is so critical and normally incorporates a number of new technologies, the risks in the selection of the correct technology and ultimately the TECHNIA – International Journal of Computing Science and Communication Technologies, VOL. 2, NO. 1, July 2009. (ISSN 0974-3375) 378 products report management can be compounded. From the survey carried out in past year ,it has realized that the general aviation industry also has limited understanding of processes for the selection of any new technology. III. JUSTIFICATION METHODS According to Meredith and Suresh (1986) investment justification methods in Advanced Manufacturing Technologies (AMT) are classified into: • economic analysis techniques • analytical methods • strategic approaches These methods deviate from each other mainly due to the treatment of non-monetary factors. Economic justification methods of manufacturing investments are discussed thoroughly by Proctor and Canada (1992). Economic analysis methods are the basic discounted cash flow techniques such as present worth, annual worth, internal rate of return and other techniques such as payback period and Return on Investment (ROI) which ignore the time value of money. The application of these techniques to the evaluation of Flexible Manufacturing System (FMS) investments is analysed by Miltenburg and Krinsky (1987). It is well known by engineering economy practitioners that accounting methods, which ignore time value of money, would produce inaccurate or at best approximate results. When flexibility, risk and non-monetary benefits are expected, and particularly if the probability distributions can be subjectively estimated, analytical procedures may be used. Strategic justification methods are qualitative in nature and are concerned with issues such as technical importance, business objectives and competitive advantage (Meredith and Suresh 1986). When strategic approaches are employed, the justification is made by considering longterm intangible benefits. Hence, using these techniques with economic or analytical methods would be more appropriate. Figure 1, which is an updated version of the classification initially proposed by Meredith and Suresh (1986), evaluates the different justification methods for AMT. Since certain criteria cannot be expressed in quantitative terms, a number of articles focus on integrating the qualitative and quantitative aspects to evaluate the benefits of an Advanced Manufacturing System (AMS). Wabalickis (1988) developed justification procedure based on the Analytic Hierarchy Process (AHP) to evaluate the numerous tangible and intangible benefits of an FMS investment. Naik and Chakravarty (1992) pointed out the need for integrating the non-financial and strategic benefits of AMS with the financial benefits and proposed a hierarchical evaluation procedure involving strategic evaluation, operational evaluation and financial evaluation. Shang and Sueyoshi (1995) proposed a selection procedure for an FMS employing the AHP, simulation and Data Envelopment Analysis (DEA). Small and Chen (1997) discussed the results of a survey conducted in the US that investigated the use of justification approaches for AMS. According to their findings, manufacturing firms using hybrid strategies, which employ both economic and strategic justification techniques, attain significantly higher levels of success from advanced technology projects. Sambasivarao and Deshmukh (1995) presented a decision support system integrating multi-attribute analysis, economic analysis and risk evaluation analysis. They suggested AHP, TOPSIS and linear additive utility model as alternative multi-attribute analysis methods. Methods include game theoretical models, multiattribute utility models, fuzzy linguistic methods and expert systems. Fig 1 : Justification methods for technology IV.FUZZY SELECTION METHODS The selection of any technology is very important to a company’s survival. As a result a large amount of research can be found in this field, with a variety of theories being presented. Karsak and Tolga (2001) proposed fuzzy decision-making as a method for evaluating new technologies. A fuzzy decision algorithm was used to select the most suitable technology considering both economic and strategic criterion. The cost or economic aspects are addressed using the fuzzy discounted cash flow analysis. Their research presented a valuable procedure that has considered a large amount of previous work in developing its findings. Karsak and Tolga (2001) looked extensively at all of this past work and proposed a fuzzy decision-making procedure as a computational-elective alternative to rectify some of the difficulties posed by the existing evaluation techniques. A paper entitled ‘Evaluation Methodologies for Technology Selection’ (Chan et al.2000), similar to the work done by Karsak and Tolga (2001), utilised Multi-Criteria DecisionKalbande and Thampi: Multi-attribute and Multi-criteria Decision Making Model for technology selection using fuzzy logic 379 Making (MCDM) in presenting a technology selection algorithm that attempted to quantify both tangible and intangible benefits within a fuzzy environment. Specifically, it described an application of the theory of fuzzy sets to hierarchical structural analysis and economic evaluations. From the analytical point of view, decisionmakers are asked to express their opinions on comparative importance of various factors in linguistic terms rather than exact numerical values. These linguistic variable scales, such as ‘very high’, ‘high’, ‘medium’, ‘low’ and ‘very low’ are then converted into fuzzy numbers, since it becomes more meaningful to quantify a subjective measurement into a range rather than in an exact value. By aggregating the hierarchy, the preferential weight of each alternative technology is found, which is called the fuzzy appropriate index. The fuzzy appropriate indices of different technologies are then ranked and preferential ranking orders of technologies are found From the economic evaluation perspective, a fuzzy cash flow analysis is employed. Since conventional engineering economic analysis involves uncertainty about future cash flows where cash flows are defined as either crisp numbers or risky probability distributions, the results of analysis may be obscure. To deal quantitatively with imprecision or uncertainty, cash flows are modelled as Triangular Fuzzy Numbers (TFN) which represent: • the most pessimistic value • the most likely possible value • the most optimistic value The algorithm presented by this paper takes the ambiguities involved in the assessment data and effectively represents and processes them to assure a more convincing and effective decision-making. V. IDENTIFICATION AND FORMULATION OF THE PROBLEM The most important part of this phase is the identification of the ultimate or the global goal as a basis for the formulation of the problem systematically. Unfortunately, in many real-life decision situations, the global goal is often ill-defined and thus cannot be straightforwardly identified. In most cases, the global goal could only be evaluated through a condensed indicator resulting from the aggregation of some criteria at the lower level. Considering this situation, all elements of the problem, such as goal(s), alternatives, constraints and the surrounded environment should be precisely identified. The information should also contain a clear description of the relationship between the listed criteria and the nature of measurement of the involved criteria. For example, for a medical service selection, the ultimate goal, assumed, could be the cure for the disease detected. Hence, the technologies required for that would be the appropriate medical equipments for the treatment and several diagnosis related elements to be used throughout the process. Based on the obtained information, a fuzzy multi criteria decision making problem can then be formulated to represent the real decision problem. VI. SUGGESTED DECISION MODEL AND FUZZIFICATION OF INPUT VARIABLE Depending on the field for application of the technologies to be selected for, list of basic questions are to be prepared and these questions need to be answered for analysis of the importance of each factor in playing the role to achieve the ultimate goal of the application. For example, some of the issues to consider when evaluating a technology, platform, language, or product are: 1. Can we achieve the desired functionality with it? 2. Does it support our performance and robustness needs? 3. Does it support our usability and/or maintainability needs? 4. What platforms will our system run on? 5. Can updates be obtained in a timely manner? 6. What is the vendor’s track record for support and enhancements? A decision model is a simplified description of a real decision situation in such a manner that allows a systematic identification of the corresponding problem and thus provides a way for the assessment of the available alternatives. One of the suitable means for describing the complex features of a decision problem is through the construction of a decision tree that provides a more clear description about the hierarchy and the logical relationship between the criteria. It is suggested that the listed attributes should be complete and exhaustive, containing mutually exclusive criteria and be restricted to performance attributes of the highest degree of importance. The technology selection aims at identifying the critical technologies which the user should concentrate its interest on, and thus, prioritizes its usage towards the final product required. The relevant technologies to be considered may result from extensive literature survey or from a panel of experts in the problem area. The fuzzification of input variables is required to facilitate the processing of the complex information into an aggregated indicator associating with the final product for information processing. In dealing with quantitative measurements involved in the evaluation process, crisp value corresponding to each criterion needs to be transformed into fuzzy form by specification of its membership functions. In this case, we have to analyze the amount features of a particular technology being used for the final product. Fuzzyfication function is introduced for each variable to express the measurement uncertainty of input variables, as more realistic approximation of the respective numerical values. TECHNIA – International Journal of Computing Science and Communication Technologies, VOL. 2, NO. 1, July 2009. (ISSN 0974-3375) 380 Conversion of fuzzy data is necessary if further information processing is to result in a global evaluation about the available alternatives. The membership function is determined and based exclusively on the decision maker’s experience or subjective preferences of the user. The critical technologies will be scored from 0 to 1 where the strength in each technology is reported as 0, if least significant and 1, if most significant. On the basis of the analysis, the importance-strength list can be constructed and sorted on the basis of the scores. In the list, the current and desired position of each critical technology is assessed. This provides the basis for formulating the technology strategy, i.e. identifying the effort needed to achieve the desired competitive position. Current Method Of Selecting New Technology Suggested Method Of Selecting New Technology VII. AGGREGATION PROCESS The aggregation operation brings the information available of all the basic technologies into a single aggregated indicator. The aggregation process reduces the original multi criteria problem into a mono criterion problem and facilitates the overall judgment about the available technological alternatives. The method of aggregation can take many forms, but it generally depends on the decision situation and the preference of the decisionmaker. The crucial point in this process is to get valid combination operators. To arrive at the final judgment on the alternatives, the existing information should be aggregated to form a condensed indicator referring to the degree to which it satisfies the global objective. In short, in this stage, all the concerned technologies are considered according to their degree of importance. One of the basic elements for the processing of fuzzy information is the employment of a fuzzy rule based inference system, comprising production rules so that it’s structure is similar to a decision support system. The information processing is generally expressed by a fuzzy rule based system, comprising two main components, namely IF portions of the statements, referred to as antecedents or premise aggregation, and THEN portions referring to the consequent aggregation results (IF a particular feature of a technology fails to process, THEN use the alternative technology’s processing). The premise aggregation consists of a combination of all the input variables into a rule to formulate the degree to which the rule is considered appropriate for the given situation. These formulations are accounted to the most-used method in formal knowledge representation, and this method is becoming interesting, due to it’s suitability, not only for experts, but also for ordinary people without a well-founded background in decision making. Combination of facts and rules about definition and identification of the key sectors of an economy, usually by applying a rule based approach gets a finite set of desired sectors. Suppose that the fuzzy set A represents a certain value of an independent variable x and B represents a value of a dependent variable y, the relationship between x and y can be defined by a set of conditional statements such as on the above statement. This definition is analogous to the definition of a non-fuzzy function f by a table of pairs ( x, f(x)) where x is a value of the argument of f and f(x) is the value of the function. The conditional statement can be represented by a fuzzy set Z if Cartesian product UA x UB with the following definition: mz(x,y)= min[mA(x),mB(y)], for every x UA and y UB Then, the grade of membership of the fuzzy value B` is given by: mB`(y) = max{min[mA(x), mz(x,y)]} x UA The fuzzy relationship between the variables is defined by conditional statements or by the acceptability statements. The technological criteria involved in the decision are the independent variables of the model and the final rating or the acceptability of the alternatives is the dependent one. If there are several independent variables involved, the general form of the fuzzy rule base system in the case of milti-input-single-output systems (MISO) is described by: IF X1 is A1 and X2 is A2, … , and Xm is Am THEN Y is BL Where (IF X1 is A1 and X2 is A2, … , and Xm is Am) are the preconditions and Y is BL refers to the post-conditions, X1, X2 and Xm are input variables, Y is the output variable. A1 is the class defined on X1, Am is the class defined on Xm and BL is class defined on Y. The antecedent or rule premise describes to what degree the rule applies, while the conclusion assigns a membership function to the output variable. The degree of support, which takes a value of between zero to one, is used to weight the corresponding statements indicating the degree of validity of the rule. If the degree of support takes a value of 1, it means that the corresponding rule is valid without restriction. On the contrary, the value of 0 reflects the full invalidity of the Kalbande and Thampi: Multi-attribute and Multi-criteria Decision Making Model for technology selection using fuzzy logic 381 respective rule. Any rule with non-zero degree of support will be taken into account. In a case, where some rules lead to a particular conclusion about a different degree of conformity, the maximum of the respective degrees of membership weighted by the degree of support will determine the final result. To reduce the complexity of the aggregation of different independent criteria, the number of criteria to be considered need to be kept as minimal as possible. If the degree of compensation takes the value 0, it reflects that there is no trade-off taken into account and accordingly, logical AND represented by min or product operator may be used. However, value of 1 indicates that the decision maker is ready to make full compensation between criteria and, thus, the logical OR represented by max operator may be used. The decision maker is generally ready to make a compensation between different criteria in which the degree of compensation may take any value in the interval of [0,11] to result in the aggregated value between those two