Skip to main content

Über dieses Buch

J. CIimaco and C. H. Antunes After the pleasure which has been to host the community of researchers and practitioners in the area of multicriteria analysis (MA) in Coimbra in August 1994, this volume of proceedings based on the papers presented at the conference is the last step of that venture. Even though this may not be the appropriate place we cannot resist, however, the temptation to express herein some brief feelings about the conference. Almost everything concerning the conference organisation has been "handcrafted" by a small number of people, with the advantages and disadvantages that this approach generates. Our first word of acknowledgement is of course due to those who have had a permanent and active role in the multiple aspects which make the success of a conference: Maria Joao Alves, Carlos Henggeler Antunes (who is a co­ author of this introduction since he has closely collaborated with me in the scientific programme), Joao Paulo Costa, Luis Dias (who greatly contributed to the organisation of this volume) and Paulo Melo, as well as Leonor Dias, from the Faculty of Economics, who has shown an outstanding dedication. To those who collaborated with the organisers in the framework of their professional activity, special thanks due to Adelina whose dedication greatly exceeded her duties. As you probably know from your own experience every small detail of the conference organisation required a lot of "sweating", but the atmosphere of joy and friendship then generated has been a generous "pay-off".



General Introduction

1. General Introduction

After the pleasure which has been to host the community of researchers and practitioners in the area of multicriteria analysis (MA) in Coimbra in August 1994, this volume of proceedings based on the papers presented at the conference is the last step of that venture. Even though this may not be the appropriate place we cannot resist, however, the temptation to express herein some brief feelings about the conference. Almost everything concerning the conference organisation has been “handcrafted” by a small number of people, with the advantages and disadvantages that this approach generates. Our first word of acknowledgement is of course due to those who have had a permanent and active role in the multiple aspects which make the success of a conference: Maria João Alves, Carlos Henggeler Antunes (who is a co-author of this introduction since he has closely collaborated with me in the scientific programme), João Paulo Costa, Luís Dias (who greatly contributed to the organisation of this volume) and Paulo Melo, as well as Leonor Dias, from the Faculty of Economics, who has shown an outstanding dedication. To those who collaborated with the organisers in the framework of their professional activity, special thanks due to Adelina whose dedication greatly exceeded her duties. As you probably know from your own experience every small detail of the conference organisation required a lot of “sweating”, but the atmosphere of joy and friendship then generated has been a generous “pay-off”.

João Clímaco, C. H. Antunes

Theory and Methodology


Complexity Issues and New Trends in Multiple Criteria Decision Aid

This paper is an introduction to the present chapter and aims to outline a framework within which all the included papers find a common conceptual thread while some terminological differences exist These reflect the complexity that is inherent to any decision aiding process in a real organisational context, a complexity which does not tolerate, for various validity reasons, a mutilating simplification that is often imposed by classical mathematical approaches. A need of new tools which might favorize a more mature and integrated development of the analyst’s activities, within cultural contexts other than those hypothesized by the classical assumptions of OR/DA, is also underlined. This permits one to recall some new trends of MCDA research which are, at the moment, found in literature.

Anna Ostanello

3. A Theoretical Framework for Measuring Attractiveness by a Categorical Based Evaluation Technique (MACBETH)

This paper presents a theoretical framework for the macbeth approach. Taking as a starting point the measurement rules used for numerically representing the qualitative (categorical) preference information assessed by the questioning procedure of macbeth, we study the existence of a solution for our particular representation problem and we briefly discuss its unicity.

Carlos A. Bana e Costa, Jean-Claude Vansnick

Interdependence In Multiple Criteria Decision Making

Modeling and optimization methods for handling multiple criteria decision problems have over the last 20 years been developed in both crisp and fuzzy environments. The overwhelming majority of approaches for finding best optimal or optimal compromise solutions to MCDM problems do not make use of the interdependences among the objectives. However, when dealing with real world decision problems we often encounter problems for which it is useful to exploit interdependences, even if it is excluded from most MCDM problem formulations. In this paper we develop and build upon measures of interdependence between the objectives, in both crisp and fuzzy formulations, and develop some methods for finding effective and more realistic solutions to MCDM problems.

Christer Carlsson, Robert Fuller

Extended Preference Structures in MultiCriteria Decision Aid

Uncertainty and ambiguity are common situations in decision aid and therefore in preference modelling and aggregation. A basic distinction between uncertainty reasons is advanced in the paper, that is between lack of information and contradictory information. A semantical investigation of this approach is conducted and the PC preference structure is adopted in order to represent the different preference situations that may occur. This type of preference structure, besides offering an enhanced granularity in preference modelling, is strongly axiomatized, enabling a precise calculus. Some of its potentialities are discussed using a preference aggregation problem.

Alexis Tsoukiàs, Philippe Vincke

6. Multiple Criteria Decision Making Within Checkland’s Soft Systems Methodology

This paper studies Checkland’s Soft Systems Methodology (SSM) as a potential multi-criteria decision making methodology. The paper first gives a brief outline of Checkland’s SSM, highlighting its salient features. It then explores in what sense it is an MCDM methodology (in contrast to an MCDM technique) for dealing with multiple-decision maker situations. In particular, we explore the role of the Weltanschauung in the MCDM context and the process of finding a compromise solution involving conflicting Weltanschauungen among multipledecision makers or stakeholders. We conclude that Checkland’s SSM can, without any changes from its usual framework, be used to bring about compromise solutions. By its very nature, it is particularly suitable for problem situations that are ill-defined and largely qualitative. For problems that lend themselves to quantification, MCDM techniques for exploring the solution space may be useful aids. The final section then explores its suitability for single-decision maker situations. The tentative conclusion reached is that, in principle, this is the case. However, the lack of a proper debate among advocates with opposing views may call for the use of MCDM techniques to get better insights into the solution space.

Hans G. Daellenbach

7. Validation Aspects of a Prototype Solution Implementation to Solve a Complex MC Problem

This paper summarises some methodological considerations concerning the development/validation process of a ’tool1 to be used as a support in a complex situation of decision making. A ’bipolar’ and multistage model is proposed for such a process, within a Multicriteria Decision Aid (MCDA) framework. This model integrates and extends two important models of modelling/validation process literature, and is conceived as a support tool for the analyst’s work during a decision aiding intervention.

Anna Ostanello

Aggregating Preferences: Outranking and Utility Function Approaches


8. Aggregating Preferences: Utility Function and Outranking Approaches

A multicriterion analysis with the aim of sheding light on a decision comprises, as any decision process, several steps. The first steps concerning the definition and structuring the decision problem take shape in a multicriterion analysis by the definition of the set A of the potential actions (feasible alternatives) and of the set A/F of the attributes or criteria from which actions are evaluated.

Jean M. Martel

9. A Concordance-Discordance Approach to Multi-Criteria Ranking of Actions with Fuzzy Evaluations

A multi-criteria ranking method based on concordance-discordance approach to construct a valued outranking relation, in the case when actions are evaluated using criteria taking fuzzy values, is presented. The use of pseudocriteria, proposed for Electre III method, is replaced by a comparison of fuzzy performances. Four different measures based on possibility and necessity concepts are applied to compare the fuzzy numbers. Then, one aggregated index is build, on which a new definition of both concordance and discordance relies. Finally, a valued outranking relation is constructed and exploited using the usual distillation procedure of Electre III. A final partial preorder results from the exploitation procedure. A practical example illustrates the method.

Poitr Czyżak, Roman Słowiński

10. A Multiplicative Method for Synthesizing Preference Ratings

Consider a discrete alternative multicriteria problem, in which the attractiveness of the alternatives is measured by absolute, ratio-scale preference ratings on a set of relevant criteria. We introduce a nonlinear, multiplicative functional form for combining the ratio-scale ratings to determine their overall priority ratings, which, depending on the preference information provided by the decision maker, can yield overall priority ratings that reflect situations of decreasing, constant or increasing marginal contributions to the achievement of the overall goal. Previous methods for combining priorities using a nonlinear functional form have either been based on interval-scale value function representations, or have assumed constant marginal contributions to the overall goal. As noted previously by other researchers, our method the overall priorities maintain the ratio-scale property, even after normalization, and precludes the occurrence of rank reversal under any condition. Our methodology includes the weighted geometric mean method for determining composite priority ratings as a special case.

Antonie Stam, A. Pedro Duarte Silva

11. Some Fundamental Problems in the Selection and Justification of Discrete Alternative MCDM Methods

The selection of the most appropriate, or "best," discrete alternative MCDM method is important in many applications. The problem of selecting the "best" MCDM method is itself an MCDM problem that does not have an obvious solution. This paper addresses several general concerns and complicated features of the selection problem. The paper argues that discrete alternative MCDM methods have inherent limitations, as the validity of a method cannot be established formally. An approach for substantiating MCDM methods is suggested, and some directions for future research are outlined.

Vladimir M. Ozernoy

The SOLVER of Volvox Multicriterion Table

VOLVOX is a consortium composed of six leading canadian firms in the field of information technology and decision making aid. The CGI group acts as manager in this project whose goal is to develop, in collaboration with several partners especially from university milieu, and to commercialize systems of decision making aid DSS in the environmental field. In addition to developing DSS responding to specific needs, the consortium intends to focus on a technological generic platform (a generic shell) VOLVOX. This shell contains several elements, one of which is the construction of a multicriterion table. The VOLVOX multicriterion table allows ordinal or cardinal, punctual or distributional evaluations. The SOLVER (multicriterion aggregation procedure) is based on the outranking synthesis approach and lies within the scope of ranking problematic. It appeals to the pseudocriteria and incomparability notions. It can also act as a negotiation and group consensus support. The purpose of this paper is to present succinctly the objectives of the VOLVOX research center as well as the main caracteristics of the solver.

Benoît Derot, Jean Gareau, Laszlo Nandor Kiss, Jean-Marc Martel

Multiobjective Mathematical Programming


Multi-objective Mathematical Programming

Although traditional mathematical programming techniques has been remarkably developed for the last half century, they do not necessarily give satisfactory results in practical applications. Because many practical problems often have several objectives conflicting with each other, while traditional mathematical programming techniques can treat only one objective function. The history of multi-objective mathematical programming began almost at the same time as nonlinear programming. In particular, it has been developed very actively since the beginning of 70’s. Now, several effective methods have been being applied to a wide range of real problems. This paper makes an introductory remarks for multi-objective mathematical programming ahead the following papers in this chapter.

Hirotaka Nakayama

On the Development and Future Aspects of Vector Optimization and MCDM

A Tutorial

The paper is divided into three parts: 1) Historical sketch, 2) Main streams of research, 3) Future development. In the first part, a concise review of the historical roots and the developments of vector optimization in the sense of decision making under several goals (objective functions) is presented. In the second part various main streams are discussed. Two basic aspects of vector optimization can be distinguished: the pure theoretical, mainly mathematical one, and the more practical, methodologically oriented one. A brief survey of vector optimization theory is given. In the third part we try to outline directions of future development based on the knowledge of past and present trends.

Tomas Gal, Thomas Hanne

Generating Interior Search Directions in Multiple Objective Linear Programming Problems Using Aspiration Levels

We propose in this paper a new approach to generating interior search directions for multiple objective linear programming (MOLP) problems. The proposed approach is based on using achievement scalarizing functions and aspiration levels to convert a problem with multiple objectives to a problem with a single objective that is amenable for a solution by the interior point affinescaling primal algorithm. The interior solution trajectory takes an interior starting solution and moves it to a nondominated solution on the efficient frontier that corresponds to the optimum of the achievement scalarizing function.

Ami Arbel, Pekka Korhonen

Preemptive Reference Point Method

Abstract. The reference point method for solving multi-criteria optimization problems is an interactive technique where the decision maker specifies requirements, similar to goal programming, in terms of aspiration levels. Ogryczak [5] showed how the reference point method could be modeled within goal programming methodology provided that the nonnegativity restrictions on weights were dropped. It allows us to consider the reference point approach as an extension of goal programming. However, in most of real-life applications of goal programming the goals are grouped according to the predefined priorities (the so-called preemptive goal programming) whereas in the reference point method all the deviations are considered to be equally important. In this paper we show how the priorities can be incorporated into the reference point method.

Włodzimierz Ogryczak

Some Optimality Conditions in Multiobjective Programming

Abstract: One of the aim of this paper is to introduce new classes of vector generalized concave functions and to point out their role in investigating local and global efficiency and in establishing sufficient optimality conditions for a vector optimization problem. Another aim is to stress the role of the Bouligand tangent cone at a point of the feasible region in deriving optimality conditions.

Alberto Cambini, Laura Martein, Riccardo Cambini

Some Remarks on Trade-off Analysis in Multi-objective Programming

In multi-objective programming, the trade-off analysis is very important because it reflects decision makers’ value judgment. In many practical applications, people usually take a scalarization of linearly weighted sum of vector-valued objective function without paying much attention to the trade-off analysis. However, the linearly weighted sum causes several difficulties in the trade-off analysis, eg. "duality gap" for nonconvex problems, and that the weight can not necessarily reflect decision makers’ value faithfully.

Hirotaka Nakayama

Eight Concepts of Optimality

The notion of optimality and the processes of optimization are clearly pivotal in the areas of economics, engineering, management and business. These are mostly "constrained" problems where search for optimality relies on some form of evaluating of tradeoffs. There can be not tradeoffs along a single dimension: optimality is therefore a fundamentally multicriterion concept. Here we develop the notion of optimum as a welldefined balance among multiple criteria. We end up with the eight different, separate and mutually irreducible optimality concepts where the traditional single-objective optimality is only a special case.

Milan Zeleny

Uncertainty in Multiobjective Programming


Interactive Fuzzy Multiobjective Programming

The paper presents an overview of interactive methods for solving three classes of fuzzy multiobjective linear programming problems: flexible programming, multiobjective linear programming with fuzzy coefficients and flexible multiobjective linear programming with fuzzy coefficients. Special attention is paid to interpretation of fuzzy goals and fuzzy constraints, to interpretation of fuzzy relations and to the character of interaction with the decision maker.

Roman Słowiński

A Fuzzy Potential Solution Approach to Multi-Criteria and Multi-Constraint Level Linear Programming Problems

Based on the decision makers1 goal-seeking and compromise behavior, this paper extends the potential solutions to fuzzy potential solutions for a given fuzzy multi-criteria and multi-constraint level (MC2) linear programming problem. We treat the MC2 problem as a problem with fuzzy optimal objectives. Then, we use the MC2-simplex method to locate two sets of potential solutions for a maximum MC2 problem and a minimum MC2 problem, respectively. With these two sets of potential solutions, we reform the ranges of the decision parameters so that each reformed range results in both the upper and lower bounds of the objective payoff over the changes of decision parameters within the range. A membership function is constructed for each of the ranges. For every reformed range, a fuzzy potential solution is determined in terms of the membership function and a primal potential basis. We also propose a heuristic algorithm to locate all fuzzy potential solutions for all possible changes of the decision parameters.

Yong Shi, Yi-Hsin Liu

Distributional Unanimity in Multiobjective Stochastic Linear Programming

Several notions of efficiency are conceivable for the multiobjective stochastic linear programming problem. In this paper, assuming that the problem’s randomness can be described by discrete scenarios with known probabilities and that decision makers’ preferences, although unknown, can be represented by a class of utility functions, we examine a set of strongly efficient solutions, the unanimous solutions. We state inclusion relations between this and other classes of efficient solutions (admissible and advocated solutions) previously studied. Under plausible assumptions about decision makers’ risk attitudes, we examine how candidates for unanimity can be generated and then tested.

F. Ben Abdelaziz, P. Lang, R. Nadeau

A Fuzzy Decomposition Method by Right-Hand-Side Allocation for Large-Scale Multiobjective Nonlinear Programming Problems

This paper deals with large-scale multiobjective nonlinear programming problems with the block angular structure in a fuzzy environment. By incorporating the fuzzy goals of the decision maker for the objective functions and adopting the add-operator for aggregating them, a fuzzy decomposition method by right-hand-side allocation is proposed. An illustrated numerical example is provided to indicate the feasibility of the proposed method.

Masatoshi Sakawa, Hitoshi Yano

Multiobjective Integer Programming


Multicriteria Integer Programming: An Overview of the Different Algorithmic Approaches

In this paper a tentative classification of different algorithmic approaches for Multicriteria Integer Programming Problems is presented. A characterization of the different approaches is also outlined.

João Clímaco, Carlos Ferreira, M. Eugénia Captivo

MOIP: A DSS For Multiple Objective Integer Programming Problems

Multiple objective integer linear programming models are often used to formulate real world problems. Our objective, in this paper, is to describe a decision support system, MOIP, designed to model and solve multiple objective integer linear programming problems interactively, efficiently, and without putting undue demands on the DM. The system is designed for IBM compatible personal computers working under DOS operating system.

V. Vassilev, S. Narula, P. Vladimirov, V. Djambov

Solving Multi-Objective Knapsack Problem by a Branch-and-Bound Procedure

A classical knapsack problem with two objectives is considered in which concepts of supported and non supported efficient solutions are pointed out.

Ulungu-Ekunda Lukata, Jacques Teghem

Bicriteria Scheduling: Minimizing Flowtime and Maximum Earliness on a Single Machine

We consider the bicriteria problem of minimizing total flowtime and maximum earliness penalties for a given set of jobs on a single machine. We discuss some properties of the efficient solutions and develop a heuristic procedure to generate all efficient solutions when inserted idle time is not allowed.

Meral Azizoglu, Suna Köksalan Kondakci, Murat Köksalan

Multicriteria Interactive Methods


Interactive Multiple Criteria Procedures: Some Reflections

This paper offers personal reflections on the evolution of interactive multiple criteria procedures. We trace their development from early roots in mathematical programming. During this early period, the theoretical and technical challenges of computing candidate solutions dominated the research. Later, major challenges arose from consideration of the behavioral aspects of decision processes. We further discuss what we consider to be current challenges and suggest some directions for future developments.

Lorraine R. Gardiner, Daniel Vanderpooten

A Method for Solving Multiple Objective Linear Programming Problems

An interactive method is presented for solving multiple objective linear programming problems. The method develops an idea for successive reduction of the set of normalized weighting coefficients. A set of Pareto efficient solutions is generated at each iteration. The dialog is in terms of aspiration levels in the objective space. A theoretical comparison with other related methods is done.

Leonid M. Kirilov, Vassil S. Vassilev

Interactive Method NIMBUS for Nondifferentiable Multiobjective Optimization Problems

An interactive method, NIMBUS, for nondifferentiable multiobjective optimization problems is introduced. The method is capable of handling several nonconvex locally Lipschitzian objective functions subject to nonlinear (possibly nondifferentiable) constraints. The idea of NIMBUS is that the decision maker can easily indicate what kind of improvements are desired and what kind of impairments are tolerable at the point considered. The decision maker is asked to classify the objective functions into five different classes: those to be improved, those to be improved down to some aspiration level, those to be accepted as they are, those to be impaired till some upper bound, and those allowed to change freely. A new problem is formed according to this classification.

Kaisa Miettinen, Marko M. Mäkelä

The LBS-Discrete Interactive Procedure for Multiple-Criteria Analysis of Decision Problems

An interactive procedure, called LBS-Discrete (Light Beam Search), for multiple-criteria analysis of decision problems with an explicitly given set of alternatives is presented. An alternative is defined as a non-dominated point in the space of criteria. The set of alternatives is assumed to be finite and relatively large. The procedure is an extension of the Light Beam Search method for linear and non-linear multiple-objective programming (Jaszkiewicz and Slowinski, 1992b) to the discrete case. While existing interactive procedures may involve the decision maker (DM) in too difficult comparisons of the candidates for the best compromise alternative, the LBS-Discrete procedure tries to overcome this inconvenience. It supports both learning about the problem and successive improvement of the current point. At the decision phase of the procedure, a sample of non-dominated points, composed of the current point and a number of alternative proposals, is presented to the DM. In order to ensure a relatively easy evaluation of the sample by the DM, it is generated in a way taking into account a preference information of intra- and inter-criteria type given by the DM with respect to the current point. The local preference model has the form of an outranking relation and defines a subregion of the non-dominated set. The sample presented to the DM comes from this subregion. The procedure can be compared to projecting a focused beam of light from a spotlight at the reference point onto the non-dominated set; the highlighted subregion changes when either the reference point or the point of interest in the non-dominated set are changed. This explains the name LBS-Discrete. Finally, a microcomputer implementation of the LBS-Discrete with an important graphical interface is characterized and its application to an agricultural problem is presented.

Andrzej Jaszkiewicz, Roman Słowiński

Group Decision and Negotiation


Support for Group Decisions and Negotiations An Overview *

This paper discusses methods for the analysis and support of group decisions and negotiations from three perspectives. First the continuation and outreach strategies are introduced; then methods and models are positioned within the modified process of negotiation proposed by Gulliver; and subsequently methods and systems developed within five areas of study are outlined.

Gregory E. Kersten

N-Person Cooperative Games with Multiple Scenarios

In this paper, solution concepts for an n-person cooperative game with multiple scenarios are considered. A characteristic function of the n -person cooperative game with multiple scenarios associates a subset of the set of all players with its real vector value. We consider the nucleolus, which is one of solution concepts based on the lexicographical framework in an n -person cooperative game, in the game with multiple scenarios. To define extended nucleoli, three aggregation methods using weighting coefficients, a minimum component, and constraints are employed. The computational methods for the three extended nucleoli are developed by repeatedly solving linear programming problems. Finally, a numerical example illustrates the proposed methods.

Masatoshi Sakawa, Ichiro Nishizaki

Successive Proposals: An Approach to the Computation and Enforcement of Efficient Agreements

Negotiations are often characterized by the fact that the participants do not know each others’ goals. This paper describes an interactive approach to computing Pareto optimal agreements under such circumstances in two-party negotiations. The approach is based on so called proposals that link the decision variables together. The proposals also provide a way of enforcing the resulting agreement, meaning that no party has an incentive to break loose from it.

Harri Ehtamo, Markku Verkama, Raimo P. Hämäläinen

Opposition and Inefficiency in Negotiated Compromises

Several studies have shown that negotiators often do not reach efficient compromises. We analyze the circumstances under which rational agents make inefficient compromises and resist improvements. We do this by describing and interpreting various negotiation situations and by developing formal consturcts and theorems for determining the character of negotiation situation. Key among these concepts is the notion of opposition. The effects of various rationality assumptions on efficiency and ther implicatios for negotiation support systems are discussed.

Gregory E. Kersten, Sunil J. Noronha

New Computer Science Developments and Multicriteria Analysis


Against Fashion: A Travel Survival Kit In “Modern” MCDA

Computer Science is going through a period of exciting new technological achievements which appear with increasing frequency. Multi-Criteria Decision Aid (MCDA), as any decision support activity, is heavily affected by such evolution, since MCDA methods are implemented and/or integrated in complex computer based information systems.

C. Antunes, A. Tsoukiàs

A Flexible Interface for Decision-Aid in Multicriteria Decision Problems

Communication between the Decision Maker (DM) and decision-aid (DA) methodologies is a main concern in Multicriteria Decision Problems, particularly when a large set of nondominated solutions exists. In a previous MCDM conference, Matos and Miranda (1989) presented the theoretical basis for a DM/DA interface (the "Anchor model"), using fuzzy sets to "translate" numerical values of the attributes into linguistic labels more meaningful to the DM. The interpretation of linguistic declaration about the DM goals was also outlined, and a new concept of "qualitative domination" was introduced.

Manual A. Matos, Pedro Borges

Decision Support for MCDM That Is Neural Network-Based and Can Learn

In this paper we consider the basic ideas, the implementation, and the application of an object-oriented decision support system (DSS), especially for multiple criteria decision making (MCDM). One main idea of the DSS is to utilize the features of different MCDM methods by integration. We discuss several approaches of integration esp. the combination of methods, the use of neural networks, and a generalization of network-like structures. Information acquisition is an essential part for any DSS. Beside aspects of interactive data entry, we discuss the utilization of historical data (especially from former decisions) for learning. By doing so the parameters of a method, some methods, or a network of methods can be tuned and we can answer questions like: Which method should be used? How can the parameters be adjusted? Is it useful to apply different methods and to aggregate their results to build a compromise solution? And how should this be done?

Thomas Hanne

Behavioral Research and Multicriteria Analysis


Behavioral Issues in MCDM: Neglected Research Questions

Behavior decision theorists have studied human decision making in great detail. Since the late 1960’s, Einhorn, Edwards, Kahneman, Roy, Trevsky, and others have developed new thoeries to explain choice and decision behavior. Thus far this behavior research has had little impact on multiple criteria decision making (MCDM). Only a handful of MCDM-research have critically examined the behavioral underpinnings of our field. To improve the success of decision tools in practice, MCDM-research should pay more attention to the behavioral realities of decision making. In this paper, we discuss various behavioral issues relevent for MCDM based on our personal observations and experiments with human subjects. The spirit of our paper is to pose questions rather than provide definite answers.

Pekka Korhonen, Jyrki Wallenius

Multicriteria Approaches in Practice — Applications and Methodological Issues



Real-world problems and decision situations generally involve the need to consider explicitly multiple,, conflicting and incommensurate evaluation aspects. This fact is generally recognized as the starting point of multiple criteria decison aid (MCDA) as a scientific discipline, although some authors consider the multiple criteria approach as essentially an "act of faith" (Bouyssou, 1993). Even though the real-world application (with the meaning: whenever someone pays for it!) of the large corpus of techniques and methods MCDA has been developed lags well behind the theoretical and methodological advances, reports on MCDA in practice are steadily growing in the scientific literature. For instance, White (1990) presented a survey listing more than 500 references covering the period 1955-86 (excluding those approches which used an a-priori explicit value function). However, one must realize that just a small number of cases actually involves an implementation phase (and most of those seems to be goal programming-based studies). The improvement of the implementation rate of MCDA studies should be a crucial concern of MCDA researchers and practitioners, namely having in mind the conceptual and operational validation of the use of MCDA techniques in real-world problems with actual decision makers (Bouyssou et al., 1993). This trend to emphasize actual studies, as well as the search for new potential areas of application, also contributes to make MCDA to receive further cross-fertilization from other scientific disciplines. Nevertheless, it is fair to recognize that real-world applications are not easy to get from organizations (both in public and private sectors) even in a free an researchoriented basis. Under these (difficult) circumstances, sound academic case-studies (clearly assumed as such) supplied with realistic data may have an important role to play, not just as valuable experimentation frameworks, but also as evangelization tools to show the potential benefits which can be harvested from an MCDA approach in complex decision situations.

João Clímaco

An Experimental Comparison of MOLP Interactive Approaches Based on a Regional Planning Model

This paper reports how previous methodological studies and experiments in power system planning and telecommunication network moderization planning led to the development of interactive environments to deal with multiple objective linear programming problemms (MOLP), particularly with three objective functions, paying special attention to the use of graphical displays and controls. These tools are well suited to assist decision maakers (DMs) in making a progressive and selective familiarization with the effieciet solution set. However, whenever the problem demand the consideration of more than three objective functions, some of those techniques could not be used, and the interactive environments had to be expanded in a coherent manner to preserve the usefulness of visual inspection as a cornerstone of computer tools to assist DMs. SOMMIX is control panel-based interactive environment which offers a large set of commmands (embodying search strategies, techniques to compute ew nondominated solutions, ways to express his/her preference, means of iteraction and information presentation, etc.) to assist the user, by facilitating and motivating the successive actions throughout the interactve solution search process. The main aim is to shed some light on the usefulness of computer packages which are flexible, user-friendly and technically sound to accomodate the very nature of human decision making processes. Using a regional planning model as a laboratory case study this paper is aimed at presentig some computer experiments with two classical multiple objectivve interactive methods (STEM and Zoints-Wallenius) and the SOMMIX ackage.

Marie J. Alves, Henggeler C. Antunes, João Clímaco

A Multiple Criteria Approach to the Estimation of Environmental Quality

This paper shows the role of multicriteria value functions in the context of environmental decision making. It focuses on the structuring of expert judgement and on the information necessary to overcome the problems related to the customary evaluation based on environmental standards. These standards, although widely used, do not provide sufficient support for the decision process and do not allow for compromises between multiple objectives. The shortcomings of the standards are highlighted, the necessary information to improve decision making is discussed and the use of value function techniques is illustrated.

Euro Beinat, Piet Rietveld

Application of MCDA to Corporate Loan Requests Evaluation

The corporate loan decision making process is a central concern in financial literature and amongst practitioners. The process, though entails many difficulties. There is, for example, no consensus in the literature on the choice or on the relative weight of the potentially relevant aspects (the macro-economic, financial, and management assessment aspects, and the aspect of the value of loan collateral) that one should take into account in a decision on a loan. The literature is also in disagreement as to how these aspects can be combined (aggregation) in order to generate a global judgement. Moreover, the information necessary in seizing these aspects most often comes in different scale units and results from evaluations which are subjective or more or less exact. Most models that are developed for loan decision, limit themselves to quantitative financial variables; they also tend to evacuate or ignore the role of the loan officer in the process, and this even if the decision maker’s judgement and appreciation of qualitative aspects is recognized as a key factor in the process. This paper shows how a multicriterion methodology based on outranking relationships can assist the loan officer in his or her attempt to surmount the difficulties involved in loan request evaluation. The method proposed is tested over a sample that contains about three hundreds of loan requests addressed to a large Canadian financial institution

Michel Bergeron, Jean-Marc Martel, Pater Twarabimenye

DSS: They Do Work!

We have heard that DSS are of little use in supporting decision-making in strategic management so many times that many of us believe it is true. We have a different story.

Malin Brännback

OpTIX-II: A Software Environment for MCDM Based on Distributed and Parallel Computing

The aim of this contribution is to provide an introduction to the OpTiX-II Software Environment. This software system supports nonsequential solution approaches for so-called multidisciplinary decision problems based on nonlinear optimization tasks (eg. [1]). The underlying mathematical programming problem can be formulated as follows: find an n-dimensional vector x = (XV ...,xn)TE RN that minimizes f ( x ) : R N R subject to the constraints g. ( J C ) = o, g . ( J C ) < o and the bounds xt<x^ xu> where gk ( J C ) :Rn –“ R (i = 1,..., me; j = m e+l , ..., m; k = 1 , m ) .

Manfred Grauer, Harald Boden

Testing a Bicriterion Location-Allocation Model with Real-World Network Traffic: The Case of Edmonton, Canada

The p-median model in a network treats demand for service as weights expressed at nodes. The flow-capturing location-allocation model treats demands expressed by traffic (OD) flows. We consider a bicriterion model, a hybrid of these two, which trades off node- and flow-based demand, and apply it to realworld data. These are journey to work data in a system of 177 traffic zones in Edmonton, Canada. The system is relatively complex, comprising 703 nodes, 2198 links and 23,350 OD pairs.

M. John Hodgson, Kenneth E. Rosing, A. Leontien G. Storrier

On The Use Of Multicriteria Decision Aid Methods To Portfolio Selection

This paper proposes the use of different multicriteria decision aid methods for management of stocks’ portfolios. The ELECTRE TRI (ELimination Et Choix Traduisant la REalit6) method and the MINORA (Multicriteria INteractive Ordinal Regression Analysis) system are used to sort and rank respectively a sample of stocks. Then, the ADELAIS (Aide k la DEcision pour systemes Lineaires multicriteres par Aide a la Structuration des preferences) multiobjective programming system is employed for the constitution of a stocks’ portfolio according to the results obtained by ELECTRE TRI and MINORA. The data used were obtained from the Athens Stock Exchange for the period 1990-91.

Ch. Hurson, C. Zopounidis

48. Interfirm Performance Evaluation Under Uncertainty: A Multi-dimensional Framework

In this paper a multidimensional framework for performance evaluation is proposed, which is elaborated and illustrated through a reallife case. A particular feature of the approach is that it takes account of the influence of risks, which are beyond the control of the decision maker. Here risk is seen as a multidemensional measure; it is expressed by means of sensitivities to unexpected changes of a multitude of risk factors. Furthermore, the approach corrects for differences in operational characteristics between the firms that are being compared. Some operational characteristics are fixed and thus uncotrollable for the decision maker, whereas other operational characteristics may be changed and controlled by the decision maker. The approach also consider the fact that it might have been better to change some of the cotrollable firm characteristics

Jaap Spronk, Erik M. Vermeulen

Scenario Analysis and Multicriteria Decision Making

The use of scenarios as internally consistent representations of possible futures is being widely used for strategic planning, in preference to attempting to forecast the future in detail. This has two implications for the application of MCDM to strategic planning problems in large corporations or in the public sector. Firstly, "policy scenarios" may be used as a representation of possible policy directions, but which are neither complete nor exhaustive. Here the challenges to MCDM relate to the selection of good scenarios, and to the provision of meaningful methods of comparisons between these scenarios. Secondly, future uncertainties will not be expressed as probability distributions, but rather as a small number of "state scenarios" (again neither exhaustive nor complete). We discuss reasons why such scenarios should be treated as part of the objectives’ hierarchy, rather than to try to collapse across them by taking expectations, for example. This, too, has implications for choice of MCDM procedures and for subsequent sensitivity studies.

Theodor J. Stewart

A Multicriteria Approach for Evaluating Efficiency and Appropriateness of Industries: A Case in China

This paper proposes a multiple criteria decision making approach to investigate the relative strengths and weaknesses of industrial zones and to analyze the appropriateness of industries for promotion. The approach is two-tiered. Firstly, it examines the industrial sector using Data Envelopment Analysis (DEA), assessing the efficiency of the decision-making units (special economic zones) with respect to certain important factors of development. For those so-called "inefficient" units, DEA is used to identify comparable "efficient" units and then set performance targets for the former vis-a-vis the latter. The findings obtained from DEA on the appropriateness of criteria are then used for developing an Analytic Hierarchy Process (AHP) model to determine the type(s) of industries suited for the conditions of the region for the improvement of the industrial sector’s efficiency. The proposed approach was applied to the case of selected special economic zones in the Southeastern part of China.

Mario T. Tabucanon, Yu Zhijing

Valuation of Environmental Goods with Expert Panels

The present work introduces the use of decision panels to valuate environmental goods for public policy purposes as a supplement to more traditional contingent evaluation methods. It addresses problems arising from road traffic and seeks to specify the most important environmental- and health impacts in a useful way, both for valuation and policy making purposes. Particularly, we want contribute with reasonable appraisals of how much it is worth for the society to reduce the various impacts by one unit.

Fred Wenstøp, Arne J. Carlsen, Olvar Bergland, Per Magnus

A Case-study in Military Decision-making With Mission-oriented Multiple Objectives

In this paper we report on a problem of selecting military equipment. We have applied an evaluation methodology based on hierarchical multi-criteria analysis, which relates the contribution of alternative options (regarding equipment, organisational structure, etc.) to military objectives by explicitly defining military missions, tasks and functions. The analysis consists of two approaches, an aggregate ("holistic") and a detailed approach, which should complement each other and offer an opportunity for comparison and reflection. We begin with an overview of the evaluation methodology and then focus on the application.

Diederik J. D. Wijnmalen

Future Trends on Multicriteria Analysis


Thoughts on the Future of the Multicriteria Field: Basic Convictions and Outline for a General Methodology

In this paper we develop some thoughts on the future of the multicriteria field. We defend the need of theoretical robustness and practical validation of the multicriteria tools, and we state two methodological convictions as pillars for the re-orientation of the future of MCDA: the conviction of the interconnection and inseparability of the objective and subjective elements of a decision context, and the conviction of constructivism and learning. Finally, we call for a collective effort to build a general methodology that can embrace the different practices and plurality of valid multicriteria methods.

Carlos A. Bana e Costa, Marc Pirlot

MCDM Trends in Natural Resources and Environmental Management

Recent events and future trends in natural resources and environmental management suggest that multiple criteria decision making (MCDM) will play an ever-more important and central role in these sectors. The emergence of ecosystem management and sustainable development, the first signs of movement away from "command and control" approaches for environmental protection, more emphasis on risk-benefit analysis, and the trend toward decentralized, consensus-based decision making form the basis for the increased need for MCDM approaches. I discuss each of these elements in turn after first defining some terms.

Jared L. Cohon

Research Agenda in MCDM

Looking in the past one can trace the first origin of MCDM in economical research related to the field of the utility theory [l].That research gave a foundation to the notion of a "rational man" making the best economical decisions. The second origin of MCDM is the field of operation research [2]. The operation research gives tools for a choice of optimal solutions in complex systemsimethods of mathematical programming[3]. The tools are very efficient in solving well structured problems. It is possible also to say that MCDM has an origin in engineering research on military technical systems[4]. Well known results of the research are the systems analysis and the method "cost-efficiency" [5].

Oleg Larichev

MCDM at the Crossroads

This paper discusses challenges and opportunities facing the multiple criteria decision making (MCDM) researchers. It reviews some of them and presents the author’s view regarding possible courses of action.

Wojtek Michalowski

Non-Fully Resolved Questions about the Efficient/Nondominated Set

In this paper is discussed the state-of-the-art in the following topics pertaining to efficient and nondominated sets in multiple objective programming: (a) number of efficient extreme points (b) software implications (c) distribution of maximally efficient facets as to their dimensionality (d) criterion cone decomposition (e) proportion of unsupported nondominated criterion vectors (f) quad-trees for storage and retrieval (g) distributions of U-valuesNotation and Terminology Employed

Ralph E. Steuer

Future Trends in MCDM

We comment firstly on the wide diversity of MCDM techniques which exist, and suggest that at least partly, the diversity arises from three different decision contexts which can be recognized. Recognition of these different contexts may assist in developing a "meta-approach" to MCDM, which is the first challenge to MCDM which is identified. Other lines of development are then suggested, including the need for alternative methods of treating risk in MCDM, for methods of visualization of the decision space, and for standards for validation of MCDM procedures. Finally, we emphasize the need for MCDM to exploit methods of soft systems and problem structuring, and to contribute to the development of groupware for group decision support.

Theodor J. Stewart

Towards the Tradeoffs-Free Optimality in MCDM

Multiple Criteria Decision Making (MCDM) has for decades relied on the traditional optimality concept of OR/MS, based on a single scalar function maximization with respect to a priori given constraints. Whenever it encountered a proper vector function, MCDM has promptly resorted to its aggregation and scalarization.

Milan Zeleny

Some Thoughts on MCDM: Myths and Ideas

Decision making is a most important function and process of management. It is one of the most pervasive problems of management and a pervasive problem in general. Multiple criteria decision making is an approach to decision making. It is the making of decisions taking into consideration multiple conflicting objectives. How do we make decisions? What is involved? There are many simple paradigms, and until recently about all that could be said was that you should list the alternatives, score them on certain dimensions, and then somehow aggregate the scores, and choose an alternative. Unfortunately, many excellent decision makers simply choose an alternative, based on instinct, intuition, or feel. Then they go on. Though that is fine in many cases; in some cases it is not. In this paper I consider some myths of decision making. I then propose some steps that we, as decision analysts and scientists, should consider taking to try to sensitize decision makers to thinking more about their decisions and making better decisions.

Stanley Zoints

The European School of MCDA : Some Recent Trends

This paper shortly describes the main characteristics of the European School of MCDA. Then, it provides some recent research directions showing the future development of this School. The paper also includes a recent bibliography on the subject of MCDA.

Constantin Zopounidis


Weitere Informationen