The Use of the Analytic Network Process to Predict the Replacement of the Gasoline Engine -Edición Única

(1)

INSTITUTO TECNOLÓGICO Y DE ESTUDIOS SUPERIORES DE MONTERREY

PRESENTE.-Por medio de la presente hago constar que soy autor y titular de la obra

denominada "THE USE OF THE ANALYTIC NETWORK PROCESS TO

PREDICT THE REPLACEMENT OF THE GASOLINE ENGINE", en los sucesivo LA OBRA, en virtud de lo cual autorizo a el Instituto Tecnológico y de Estudios

Superiores de Monterrey (EL INSTITUTO) para que efectúe la divulgación,

publicación, comunicación pública, distribución, distribución pública y

reproducción, así como la digitalización de la misma, con fines académicos o

propios al objeto de EL INSTITUTO, dentro del círculo de la comunidad del

Tecnológico de Monterrey.

El Instituto se compromete a respetar en todo momento mi autoría y a

otorgarme el crédito correspondiente en todas las actividades mencionadas

anteriormente de la obra.

De la misma manera, manifiesto que el contenido académico, literario, la

edición y en general cualquier parte de LA OBRA son de mi entera

responsabilidad, por lo que deslindo a EL INSTITUTO por cualquier violación a

los derechos de autor y/o propiedad intelectual y/o cualquier responsabilidad

(2)

The Use of the Analytic Network Process to Predict the

Replacement of the Gasoline Engine -Edición Única

Title The Use of the Analytic Network Process to Predict the Replacement of the Gasoline Engine -Edición Única

Authors Javier Eduardo Niño Rodríguez

Affiliation Tecnológico de Monterrey, Campus Monterrey

Issue Date 2009-12-01

Item type Tesis

Rights Open Access

Downloaded 18-Jan-2017 23:39:20

(3)

INSTITUTO TECNOLÓGICO Y DE ESTUDIOS SUPERIORES DE MONTERREY

CAMPUS MONTERREY

DIVISIÓN DE INGENIERÍA Y ARQUITECTURA PROGRAMA DE GRADUADOS EN INGENIERÍA

THE USE OF THE ANALYTIC NETWORK PROCESS TO PREDICT THE REPLACEMENT OF THE GASOLINE ENGINE.

TESIS

PRESENTADA COMO REQUISITO PARCIAL PARA OBTENER EL GRADO ACADEMICO DE:

MAESTRO EN CIENCIAS

CON ESPECIALIDAD EN SISTEMAS DE CALIDAD Y PRODUCTIVIDAD

POR:

JAVIER EDUARDO NIÑO RODRÍGUEZ

(4)

i INSTITUTO TECNOLÓGICO Y DE ESTUDIOS SUPERIORES DE MONTERREY

CAMPUS MONTERREY

DIVISIÓN DE INGENIERÍA Y ARQUITECTURA PROGRAMA DE GRADUADOS EN INGENIERÍA

Los miembros del comité de tesis recomendamos que el presente proyecto de tesis

presentado por el Ing. Javier Eduardo Niño Rodríguez sea aceptado como requisito

parcial para obtener el grado académico de:

Maestro en Ciencias

Con Especialidad Sistemas de Calidad y Productividad

Comité de Tesis:

_________________________ Dr. David Güemes Castorena

Asesor

_______________________ ___________________

Dr. Neale Ricardo Smith Cornejo Dr. Noel León Rovira

Sinodal Sinodal

Aprobado:

_______________________

Dr. Neale Ricardo Smith Cornejo

Director de Maestría en Ciencias con Especialidad en Sistemas de Calidad y Productividad.

(5)

ii

Dedicatoria

A Dios, quien me ha dado

una vida maravillosa llena

de oportunidades.

A mi Mamá, quien me ha

enseñado perseverancia y

coraje para seguir adelante

y quien siempre ha creído en

mi.

A mi Papá, quien me ha

enseñado las pequeñas pero

importantes cosas que no

están en los libros.

A mi Novia Gabriela, quien

ha sido mi apoyo y mi

compañía incondicional.

A mi familia, quien siempre

me ha dado su confianza y

apoyo.

(6)

iii

Agradecimientos

A mi asesor Dr. David Güemes, quien confió en mí, me brindó su apoyo y orientación para el desarrollo de esta investigación.

Al Dr. Noel León y al Dr. Neale Smith, por su apoyo y contribuciones para enriquecer los resultados de esta investigación.

A la Cátedra de Estudios Prospectivos del Tecnológico de Monterrey, quien me apoyó con una beca para el estudio de esta maestría.

(7)

iv

Abstract

The opinion of a single person in decision making can become insufficient when analyzing complex problems, especially those whose solutions can affect many others. Due to the above a discussion and interchange between the actors involved, which by their experience and knowledge can help to structure the problem and evaluate possible solutions, should be facilitate. The conflict arises when the decisions taken have not sustentation and the resolution is subject to the subjectivity and creativity of those involved, which can lead to inappropriate ways to ignore certain aspects of the problematic situation among other unwanted scenarios in the process of group decision making. Another important aspect to consider is foresight (a future vision), as it has gained importance in recent years. Nowadays it is common to find the appearance of foresight studies on topics such as technology, conflict resolution, regional development, or national and international economic dynamics.

In the process of group decision making, generally we use qualitative interventions (methods that give meaning to events and perceptions). And what is sought for study and analysis of this thesis is the use of the Analytic Network Process (ANP), a technique that use qualitative pair wise with quantitative methods (measured variables and methods that apply analysis using or generating reliable and valid). This in a foresight environment that will allow decisions to have sustentation based on mathematics and in this way to reduce subjectivity.

(8)

v

List of Figures

Figure . Qualitative vs. Quantitative. Adapted from L. Georghiou ...

[image:10.612.85.530.110.342.2]

(11)

viii

List of Tables

Table . The Fundamental Scale. Saaty & Vargas, ... Table 2. AHP Example (Coyle, 2004) ... Table 3. CI value from large samples of matrices. (Coyle, 2004) ... Table . Pollution pair wise comparison. ... Table . Green Tech For Creation pair wise comparison. ... Table . Fuel Cost pair wise comparison...

Table .Average Fuel Prices US Department on Energy, ...

Table . Price pair wise comparison. ... Table . Savings pair wise comparison. ... Table . Tax )ncentives pair wise comparison. ... Table . Availability pair wise comparison. ... Table . )nfrastructure pair wise comparison. ... Table . Reputation pair wise comparison. ... Table . ‐ pair wise comparison. ... Table . ‐ pair wise comparison. ... Table . ‐ pair wise comparison. ... Table . ‐ pair wise comparison. ... Table . Pollution pair wise comparison. ... Table . Green Tech For Creation pair wise comparison. ... Table . Fuel Cost pair wise comparison. ... Table . Price pair wise comparison. ... Table . Savings pair wise comparison. ... Table . Tax )ncentives pair wise comparison. ... Table . Availability pair wise comparison. ... Table . )nfrastructure pair wise comparison. ... Table . Reputation pair wise comparison. ...

Table . Pollution pair wise comparison to calculate the eigenvector. ...

Table . Pollution pair wise comparison. Eigenvector example ...

Table . Pollution pair wise comparison. Normalized matrix. ...

(12)

9

Chapter 1 Introduction

There are two known ways to analyze causal influences and their effects. One is by using traditional deductive logic beginning with assumptions and carefully deducing an outcome from them. The other is a holistic approach in which all the factors and criteria involved are laid out in advance in a hierarchy or in a network system that allows for dependencies (Saaty & Vargas, 2006).

The first way is a linear approach in which several separate conclusions may be obtained and the problem is to piece them together in some coherent way which needs imagination and experience as logic tells us little or nothing about how to bring the different conclusions into an integrated outcome. In the second approach all possible outcomes that can be thought of are joined together in these structures and the both judgment and logic are use to estimate the relative influence from which the overall answer is derived. “This approach requires knowledge and experience with the subject, and is not totally dependent on the ability to reason logically which most people cannot do well anyway and which is not guaranteed to discover the truth because the assumptions may be poor and the reasoning faulty” (Saaty & Vargas, 2006).

People who work in decision making have to deal with the measure of quantitative and qualitative attributes. “The question is whether there is a coherent theory that can deal with both these worlds of reality without compromising either. The Analytic Network Process (ANP) is a method that can be used to establish measures in both domains” (Saaty & Vargas, 2006).

(13)

10

Chapter 1 The Use of the ANP to predict the Replacement of the Gasoline Engine

action, are factors that contribute to the difficulty of making effective decisions. Due to these reasons, the decision-makers need to use any help at hand was never more apparent.

1.1. Background of Concepts.

Modern decision-making has its origins in the development of operations research during World War II. Investigations during times of war carried out by multidisciplinary teams were initiated in Britain with significant contributions from United States. "The operations research was developed to analyze the quantitative aspects of repetitive situations. Then possible solutions were proposed, analyzed, tested, evaluated and modified when necessary at the time of its implementation " (Trueman, 1981).

"At the end of the war. An effort was made to apply the methodology of operations research to problems of business and industry " (Trueman, 1981). In 1947 George Dantzing was responsible for developing the simplex algorithm for solving linear programming problems, which has undoubtedly been the most widely used algorithm for solving problems through the development of computers. In 1950 there was steady growth in non-military applications of operations research in tandem with a strong interest in professional development and education in this field. In 1954 Leslie Edie was the first to successfully use the theory of queues in a problem of the industry, which was previously only an application for telephone companies. In 1958 developed in the United States Navy appears the Critical Path Methodology CPM. "During the 60's many operation research teams were involved in a large number of companies, often reporting to more senior positions within the same" (Trueman, 1981).

(14)

11

information by combining these methods. Unlike publications that focus on research, group dynamics, organizational change and international administration tend to receive more manuscripts using both methods. Another change that has occurred in decision making is that decisions have been made by groups not by individuals.

Proposed by Thomas Saaty in 1980. The Analytic Hierarchy Process AHP is a generic problem-solving approach that is used in making complex multi-criteria decisions based on variables that do not have exact numerical consequences. The decision problem is represented in the form of a hierarchical structure, with the apex being the overall focus or objective, criteria at the middle and the decision alternatives at the bottom. Such a configuration represents the basic three-level model of AHP. Nevertheless, several levels like sub goals, sub criteria, scenarios etc. could be considered in the model depending on the construction of the decision problem (Sureshchandar & Leisten, 2006).

The Analytic Network Process ANP is a multicriteria theory of measurement used to derive relative priority scales of absolutes numbers from individual judgments that also belong to a fundamental scale of absolute numbers. These judgments represent the relative influence, of one or two elements over the other in a pair wise comparison process on a third element in the system, with respect to an underlying control criterion. Through its super matrix whose entries are themselves matrices of column priorities, the ANP synthesizes the outcome of dependence and feedback within and between clusters of elements. The Analytic Hierarchy Process (AHP) with its independence assumptions on upper levels from lower levels and the independence of elements in a lever is a special case of the ANP (Saaty, 2005).

(15)

12

1.1.1 Background of the problem.

The idea of combining qualitative and quantitative approaches in one study owes much to discussions of the past on the idea of mixing methods, to link the methods and paradigms and combine research designs. In all phases of study, in terms of mixing methods in 1959 Campbell and Fisk saw the use of more than one method to measure the psychological treatment to ensure that variance was reflected in the treatment and not the method. Denzin in 1978 use the term triangulation used to argue about the combination of methodologies in the study of the same phenomenon. The concept of triangulation is based on the premise that the method, information sources and researchers will be neutralized1 when used in conjunction with other information, investigators and methods (Creswell, 1994). A combined method of study is one in which the researcher uses multiple methods of data collection and analysis. These methods can be approached from “inside the methods" such as different types of quantitative strategies for collecting information and also involves "between methods" that lies in procedures for collecting quantitative and qualitative information (Creswell, 1994).

In the mixed methods methodology, 3 models emerge from the literature (Creswell, 1994). 1. 2 phases Design: Where there is a qualitative study first, then a quantitative study.

2. Less-dominant, dominant Design: In this design the researcher present a study based mainly on a dominant paradigm, with a small component of the study complemented with a minor paradigm.

3. Mixed methodology Design: This design provides the highest combination of paradigms; here the researcher presents a mix of qualitative and quantitative aspects through every step of the investigation.

For this research the design 3 which has mixed aspects of qualitative and quantitative methods throughout the investigation is going to be used.

1

(16)

13

"To work properly in the field, decision makers must not only possess tactical skills, but also a good measure of "wisdom" or tacit knowledge to deal with people, policies, organizations, etc. To avoid compromising the effectiveness of decisions made by consensus " (Trueman, 1981).

1.1.1.1 Background of fuel industry.

The Oil and Gas Journal (OGJ) estimates that at the beginning of 2004, world-wide reserves of petroleum was 1.27 trillion barrels. This estimate is 53 billion barrels higher than prior year (2003), which reflected additional discoveries, improving technology, and changing economics. If we use the world petroleum consumption rate of 2003 as a fixed rate, the worldwide petroleum reserve would be able to sustain the current level of consumption for additional 43.4 years (Sunggyu, Speight, & Loyalka, 2007).

Currently, transportation, fuel and petrochemical industries depend heavily upon petroleum-based feedstocks. Therefore, alternative fuels replacing petrochemical feedstocks and supplementing petroleum derived materials must be developed and utilize more. Necessary infrastructure also needs to be developed and changed to make a transition from current petroleum economy (Sunggyu, Speight, & Loyalka, 2007).

The unprecedented popularity and successful utilization of petroleum resources observed in the 20th century may have to decline in the 21st century owing a lack of resource availability, thus making prospects for future sustainability seem grim (Sunggyu, Speight, & Loyalka, 2007).

During the past several decades, there has been considerable increase in research and development in areas or environmentally acceptable alternative fuels. Efforts have been made to reduce emissions of air pollutants associated with combustion processes whose sources include electric power generation and vehicular transportation. Air pollutants that have been targeted for minimization or elimination include SOx, NOx, COx, VOCs,

(17)

14

The use of alternative fuels is no longer a matter for the future; it is a realistic issue of the present (Sunggyu, Speight, & Loyalka, 2007).

1.1.2 Problem Statement.

The question which this research seeks to answer is: Which technology is going to replace the gasoline engine using an ANP approach to predict it? And when is this replacement will take place?

1.3. Objectives.

1.3.1 General Objective.

The development of an ANP model that combines 2 methods (qualitative and quantitative) on a foresight environment that help to predict the emergence of a technology over a group of technologies that will be replacing the gasoline engine.

1.3.2 Specific Objectives.

To help decision makers on participative process by making use of a mixed method methodology (ANP).

To support the decision taken with mathematical analysis and by this reducing the subjectivity of the participants.

To develop a systemic vision on the decision makers on the decision making process.

1.4. Research Questions.

• Can the combination of qualitative and quantitative methods using ANP help to make better decisions?

(18)

15

• Can prediction be made using the ANP approach?

1.5. Justification.

[image:18.612.218.395.232.371.2]

Luke Georghiou, on his book “The Handbook for Technology Foresight” suggests that if a general rule must exist, it will be that the best approach is one that combines both approaches qualitative and quantitative, and this is depicted in Figure 1 taken from a French mathematician named Rene Thom in 1975.

Figure 1. Qualitative vs. Quantitative. Adapted from L. Georghiou (2008)

If we assume that a particular event E has resulted in a chart like the one shown in Figure 1. To explain the event and will have 2 theories T1 and T2, which gives us the graphs g1 and g2 respectively. As you can see none of the theories that explain the event meets the graph of the event E to perfection. The graph g1 fits in a quantitative way in the sense that given the range of observation, the difference between prediction and observation is smaller than the theory g2, i.e. | | | |. But you will notice that the chart g2 has the same form and appearance that E, in a qualitative sense (Georghiou, Cassingena-Harper, Keenan, Miles, & Propper, 2008).

(19)

16

By using different methods at different points in the evaluation of an option, the assessment team can enhance the evaluation of information and minimize the weaknesses of using a method on an individual basis. "A multi-method approach can increase the validity and reliability of evaluation" (Freechtling & Sharp, 1997).

"There is a growing consensus among evaluation experts that both qualitative and quantitative methods have a place in the performance and effectiveness of evaluations. Both formative and summative assessments are enriched by a mixed methods approach ( (Freechtling & Sharp, 1997).

The ultimate goal of any assessment project is to answer the questions that were raised at first. Mixed methods (qualitative and quantitative) are useful for providing better opportunities to answer those questions (Freechtling & Sharp, 1997).

There are 3 areas, in which mixed method methodologies are superior to individual methods, this according to Abbas Tashakkori and Charles Teddle in his book "Handbook of Mixed Methods in Social & Behavioral Research”:

1. The mixed methodology can answer questions that others do not. 2. The mixed methodology provides better and stronger inferences. 3. The mixed methodology provides an opportunity to present more diversity of points of view.

(20)

17

1.6. Research Context.

This is research is limited to use of the Analytic Network Process (ANP) in order to Predict the outcome, this ANP methodology allow the use of quantitative and qualitative methodologies in order to generate results.

1.7. Variables Definition.

• Hybrid Technology

Any vehicle that has more than one power source can be classified as a hybrid, but most frequently the term is used for a vehicle which combines electric drive with a heat engine using a fossil-fuel energy source named Hybrid Electric Vehicle (Westbrook, 2001).

(21)

18

capturing the excess energy of the ICE during coasting (Chan, The State of the Art of Electric,Hybrid, and Fuel Cell Vehicles, 2007).

• Ethanol Technology.

Ethanol is a liquid fuel that contains about two-thirds of the energy value of a comparable amount of gasoline. It is commonly blended in the United States as E-IO (10 percent ethanol, 90 percent gasoline) because no modification of the engines of most cars is needed. However, the ethanol industry is pushing to have more automobile engines capable of using E-85 (85 percent ethanol, 15 percent gasoline), and some states are mandating a mixture greater than 10 percent. Ethanol can't be shipped together with gasoline in pipelines because it separates from the mixture when moisture is present, so it must be trucked to where it will be mixed with gasoline. Ethanol is produced commercially by fermenting the sugar from high-sugar crops (especially sugarcane) or by converting the starch in crops such as corn and cassava into sugars and then fermenting the sugars. The conversion of starch to sugar is fairly simple, but it is still much more costly to produce ethanol from high-starch plants than from high-sugar plants (Magdoff F. , 2008).

• Diesel Technology.

(22)

19

• Hydrogen Technology.

A hydrogen vehicle is a vehicle that uses hydrogen as its onboard fuel for motive power. The term may refer to a personal transportation vehicle, such as an automobile, or any other vehicle that uses hydrogen in a similar fashion, such as an aircraft. The power plants of such vehicles convert the chemical energy of hydrogen to mechanical energy (torque) in one of two methods: combustion, or electrochemical conversion in a fuel-cell. Hydrogen is used in a fuel-cell to produce electricity which then feeds a highly efficient electric motor driving the wheels directly (Westbrook, 2001).

Nowadays there is no fuel cell car available for sale. Automakers say they are evaluating their test fleets. GM has loaned 100 fuel cell Equinox crossovers to politicians, celebrities and a select group of the general public. Ford Motor Group operates 30 fuel cell-equipped Focus vehicles throughout the U.S., Canada and Europe, and Honda plans to lease 200 Claritys over a three-year period. Each lease runs for three years, he said, and costs $600 a month, including collision insurance and maintenance (Tablac, 2008).

Sixty-two operational hydrogen-filling stations exist in the United States, mostly on the West and East coasts, according to the National Hydrogen Association in Washington. By comparison, there were about 117,000 gasoline stations in 2007, according to U.S. Census data. GM and Shell say an infrastructure investment would be relatively small. A national network of 12,000 hydrogen stations could be built for $10 billion to $15 billion, according to a 2007 study by the two companies. Other studies call for a larger investment (Tablac, 2008).

(23)

20

Gasoline or petrol is a petroleum-derived liquid mixture, primarily used as fuel in internal combustion engines. The internal combustion engine is an engine in which the combustion of a fuel occurs with an oxidizer (usually air) in a combustion chamber. In an internal combustion engine the expansion of the high temperature and pressure gases, which are produced by the combustion, directly applies force to a movable component of the engine, such as the pistons or turbine blades and by moving it over a distance, generate useful mechanical energy (Hogarty, Gasoline: Still powering cars in 2050?, 1999).

The gasoline option is been evaluated due that may be a probability that none of the above will replace that technology and we will continue using it. The results of this study are applicable for the American Continent due that the behavior that is been seen on Europe and Asia is too different so the outcome may be different for those continents. The model is self-made and it uses the ANP rules to reach to the outcome which is to predict the technology and the time in which that technology is going to replace the gasoline engine.

1.8. Scope and Limitations of the Study.

The problem is bounded to the evaluation of five different options among them we have technologies and type of engines which are the ones evaluated to have the probability to replace the gasoline engine:

• Hybrid Technology (Gasoline and Diesel).

• Ethanol Technology.

• Diesel Technology.

• Hydrogen Technology.

• Gasoline Technology.

We can see that the limitation and use of these options have sustentation on a US Department of energy web2site where the Alternative Fuels and Advanced Vehicles Data Center state these options as alternatives to gasoline engine. Also on a IEEE publication in

2

(24)

21

2007 named: The State of the Art of Electric, Hybrid, and Fuel Cell Vehicles by C. C. Chan as we can see it on figure # 2. The Chrysler website 3 also uses these options as alternatives and innovations to the gasoline engine.

Figure 2. Road Map of Auto Technologies. (Chan, 2007)

This research is also limited to the evaluation of the technologies and engines on light-duty vehicles (automobiles and light trucks), and to the use of the ANP methodology to make the prediction.

3

(25)

21

Chapter 2 Theoretical Framework

[image:25.612.87.533.198.463.2]

2.1 Conceptual Map.

Figure 3. Conceptual Map.

2.2 Theoretical Framework

2.2.1 Group Decision Making

(26)

22 decision is deciding what action a group should take, there are various systems designed to solve this problem (Flyvbjerg, 2006):

• Consensus decision-making: tries to avoid "winners" and "losers". Consensus requires that a majority approve a given course of action, but that the minority agrees to go along with the course of action. In other words, if the minority opposes the course of action, consensus requires that the course of action be modified to remove objectionable features.

• Voting based methods

• Range voting: lets each member score one or more of the available options. The option with the highest average is chosen. This method has experimentally been shown to produce the lowest Bayesian regret among common voting methods, even when voters are strategic. The ANP is based on this voting method.

• Majority: requires support from more than 50% of the members of the group. Thus, the bar for action is lower than with unanimity and a group of “losers” is implicit to this rule.

• Plurality: where the largest block in a group decides, even if it falls short of a majority.

• Dictatorship: where one individual determines the course of action.

Currently, many decisions in organizations are made by groups or committees. There are permanent committees of executives who meet regularly, working groups specially formed to discuss specific issues, groups of projects working on creating new products and quality circles, etc. "The responsibility for making a decision usually rests in the executive, but it is clear that, he doesn’t really makes the decision without help" (Moody, 1991).

(27)

23 The main advantage of group decision making is the fact that a group can bring a broader experience and a wide variety of views. "If a problem is large, then the areas associated with the solution will cover an area greater than the experience of one person could cover" (Moody, 1991).

2.2.2 Qualitative Methods

Are methods that give meaning to events and perceptions. Such interpretations tend to be based on subjectivity and creativity that is difficult to corroborate (Moody, 1991) .

For decision making is required to collect information on a given problem and we use different techniques. "It is possible to consult experts in the specific field and rely on their experience and then make a decision" (Moody, 1991).

According to Moody (1991) there are several techniques to prove the veracity of the information received, here I present the Delphi technique, which is the one used in this research.

The Delphi technique.

It is a method to predict the future using experts in the area which belongs to the problem. It comprises a group of experts in the specific field and then they, in an independent way, they predict the future. Each member is distributed a series of questions related to their area of expertise.

"The technique has the advantage to the decision-maker to know the answers that the experts in the field given to the various questions and lost little time in gathering the answers (Moody, 1991).

2.2.3 Quantitative Methods

We are surrounded by numbers. In typical day we may find that the temperature is 17 ⁰ C, the oil cost $ 50 per barrel, 200 million unemployed people in the world. The numbers are

(28)

24 so common that we all need to use them. Quantitative methods are methods that measure variables and use or apply statistical analysis to generate reliable and valid.

"Whether in a factory, farm or a kitchen, resources as men, machines and money have to be coordinated against constraints of time and space to achieve the objectives outlined in the most efficient manner" (Srivastava, Shenoy, & Sharma, 1989).

“The main advantage of the numbers is that they give an objective measure. When we measure something and express it in numbers, we can describe exactly” (Trueman, 1981) The use of numbers increases our understanding of a situation, this does not mean that we must be experts in math to understand things, just means we should be able to appreciate the quantitative arguments and make numerical analysis (Srivastava, Shenoy, & Sharma, 1989).

“The understanding of the applicability of quantitative methods in decision making is essential for managers” (Trueman, 1981).

The forecast is an unavoidable responsibility of management. Facing uncertainty about the future, management sees the past conduct as an indicator of what is to come (Srivastava, Shenoy, & Sharma, 1989).

"Managers make decisions based on information or data available … Only if such data is properly collected, analyzed and presented will be useful for decision making" (Srivastava, Shenoy, & Sharma, 1989).

2.2.4 The Analytic Hierarchy Process (AHP).

Proposed by Thomas Saaty in 1980. The AHP is a generic problem-solving approach that is used in making complex multi-criteria decisions based on variables that do not have exact numerical consequences. The decision problem is represented in the form of a hierarchical structure like figure 1 with the apex being the overall focus or objective, criteria at the middle and the decision alternatives at the bottom. Such a configuration represents the basic three-level model of AHP. Nevertheless, several levels like sub goals, sub criteria,

(29)

25 scenarios etc. could be considered in the model depending on the construction of the decision problem (Sureshchandar & Leisten, 2006).

Figure 4. Hierarchical Structure. Adapted from Sureshchandar & Leisten, 2006.

It employs a qualitative methodology to decompose an unstructured problem into a systematic decision hierarchy. In the quantitative sense, it adopts a pair wise comparison to execute the consistency test to validate the consistency of responses. In short, AHP is a hierarchical representation of a system. A hierarchy is an abstraction of the structure of the system, consisting of several levels representing the decomposition of the overall objective to a set of clusters, sub-clusters, and so on down to the final level. Decomposing the complexity of a problem into different levels or components and synthesizing the relations of the components are the underlying concepts of AHP (Sureshchandar & Leisten, 2006).

The AHP addresses how to determine the relative importance of a set of alternatives in a multi-criteria decision problem. Since its initial development, the AHP has been applied in a wide variety of decision areas, including those related to production and operations management. It makes it possible to incorporate judgments on intangible qualitative criteria alongside tangible quantitative criteria. Elements in each level are compared pair wise with respect to their importance to an element in the next higher level and, starting at the top of the hierarchy and working down, a number of square matrices called preference matrices are created in the process of comparing elements at a given level (Sureshchandar & Leisten, 2006).

(30)

26 The normalized principal Eigen vector of the resulting comparison matrix provides a measure of each criterion’s relative importance with respect to the overall objective. Similarly, pair wise comparisons are made at the next level, the alternatives, with respect to each criterion and the corresponding pair wise comparison matrices are obtained (Sureshchandar & Leisten, 2006). The scale used in this process is the Fundamental Scale shown in Table 1, propose by Saaty. This scale has been derived through stimulus response theory and validated for effectiveness, not only in many applications by a number of people, but also through theoretical justification of what scale one must use in the comparison of elements (Saaty & Vargas, Decision Making With The Analytic Network Process, 2006). To synthesize the results over all levels, the priorities at each level are weighted by the priority of the higher level criterion with respect to which comparisons are made. The weighted priorities of the decision alternatives are added component wise in order to obtain an overall weight or priority of each alternative over the entire hierarchy (Sureshchandar & Leisten, 2006).

Intensity of

Importance Definition Explanation

1 Equal Importance Two activities contribute equally to the objective 2 Weak or slight

3 Moderate importance Experience and judgment slightly favor one activity over another

4 Moderate plus

5 Strong importance Experience and judgment strongly favor one activity over another 6 Strong plus

7 Very strong or _{demonstrated importance} An activity is favored very strongly over another; its dominance _{demonstrated in practice}

8 Very, very strong

9 Extreme importance The evidence favoring one activity over another is of the highest possible order of affirmation

Reciprocals of above

If activity i has one of the above nonzero numbers assigned to it when compared with activity j, then j has the reciprocal value when compared with i

A reasonable assumption

Rationals Ratios arising from the scale

[image:30.612.92.521.388.626.2]

If consistency were to be forced by obtaining n numerical values to span the matrix

Table 1. The Fundamental Scale. (Saaty & Vargas, 2006)

The increased acceptance of the technique among practitioners can be attributed to the helpfulness of the hierarchical problem representations and the appeal of pair-wise comparisons in preference elicitation (Sureshchandar & Leisten, 2006). At present, popular

(31)

27 software products such as Expert Choice continue to promote the AHP as an easy method of multi-criteria decision analysis with the result the range of reported practical applications is extensive. It has become quite popular in research as well due to the fact that its utility outweighs other research methods. As the methodological procedure of AHP can easily be incorporated into multiple, objective programming formulations with interactive solution process, it has received a wider attention in various fields (Sureshchandar & Leisten, 2006).

2.2.4.1 The AHP Mathematics.

Consider n elements to be compared, C1 … Cn and denote the relative ‘weight’ (or priority

or significance) of Ci with respect to Cj by aij and form a square matrix A=(aij) of order n

with the constraints that aij = 1/aji, for i ≠ j, and aii = 1, all i. Such a matrix is said to be a

reciprocal matrix (Coyle, 2004).

The weights are consistent if they are transitive, that is aik = aijajk for all i, j, and k. Such a

matrix might exist if the aij are calculated from exactly measured data. Then find a vector ω of order n such that Aω = λω . For such a matrix, ω is said to be an eigenvector (of order n) and λ is an eigenvalue. For a consistent matrix, λ = n . For matrices involving human judgment, the condition aik = aijajk does not hold as human judgments are inconsistent to a

greater or lesser degree. In such a case the ω vector satisfies the equation Aω= λmaxω and

λmax ≥ n. The difference, if any, between λmax and n is an indication of the inconsistency of

the judgments. If λmax = n then the judgments have turned out to be consistent (Coyle,

2004).

Finally, a Consistency Index can be calculated from (λmax-n)/(n-1). That needs to be

assessed against judgments made completely at random and Saaty has calculated large samples of random matrices of increasing order and the Consistency Indices of those matrices. A true Consistency Ratio is calculated by dividing the Consistency Index for the set of judgments by the Index for the corresponding random matrix. Saaty suggests that if that ratio exceeds 0.1 the set of judgments may be too inconsistent to be reliable. In practice, CRs of more than 0.1 sometimes have to be accepted. A CR of 0 means that the judgments are perfectly consistent (Coyle, 2004).

(32)

28 2.2.4.2 The AHP Calculations.

There are several methods for calculating the eigenvector. Multiplying together the entries in each row of the matrix and then taking the nth root of that product gives a very good

approximation to the correct answer. The nth roots are summed and that sum is used to

normalize the eigenvector elements to add to 1.00. In the matrix below, the 4th_{root for the} first row is 0.293 and that is divided by 5.024 to give 0.058 as the first element in the eigenvector. The table below gives a worked example in terms of four attributes to be compared which, for simplicity, we refer to as A, B, C, and D (Coyle, 2004).

Table 2. AHP Example (Coyle, 2004)

The eigenvector of the relative importance or value of A, B, C and D is (0.058,0.262,0.454,0.226). Thus, C is the most valuable, B and D are behind, but roughly equal and A is very much less significant.

2.2.4.3 Consistency.

The next stage is to calculate λmax so as to lead to the Consistency Index and the

Consistency Ratio. We first multiply on the right the matrix of judgments by the eigenvector, obtaining a new vector. The calculation for the first row in the matrix is: 1*0.058+1/3*0.262+1/9*0.454+1/5*0.226 = 0.240 and the remaining three rows give 1.116, 1.916 and 0.928. This vector of four elements (0.240,1.116,1.916,0.928) is, of course, the product Aω and the AHP theory says that Aω=λmaxω so we can now get four

estimates of λmax by the simple expedient of dividing each component of

(0.240,1.116,1.916,0.928) by the corresponding eigenvector element. This gives 0.240/0.058=4.137 together with 4.259, 4.22 and 4.11. The mean of these values is 4.18

(33)

29 and that is our estimate for λmax. If any of the estimates for λmax turns out to be less than n,

or 4 in this case, there has been an error in the calculation, which is a useful sanity check (Coyle, 2004).

The Consistency Index for a matrix is calculated from (λmax-n)/(n-1) and, since n=4 for this

matrix, the CI is 0.060. The final step is to calculate the Consistency Ratio for this set of judgments using the CI for the corresponding value from large samples of matrices of purely random judgments using the table below, derived from Saaty’s book, in which the upper row is the order of the random matrix, and the lower is the corresponding index of consistency for random judgments (Coyle, 2004).

Table 3. CI value from large samples of matrices. (Coyle, 2004)

For this example, that gives 0.060/0.90=0.0677. Saaty argues that a CR > 0.1 indicates that the judgments are at the limit of consistency though CRs > 0.1 have to be accepted sometimes. In this instance, we are on safe ground. A CR as high as, say, 0.9 would mean that the pair wise judgments are just about random and are completely untrustworthy (Coyle, 2004).

AHP considers both qualitative and quantitative approaches to research and combines them into a single empirical inquiry. It is to be noted that the AHP approach is a subjective methodology that does not necessarily involve a large number of experts to take part in the process. Certainly, in an academic research, a small sample might only provide a very rough picture. Nonetheless, with reference to important business decisions, opinions from a small group of key executives of the company are sufficient for generating reliable and useful results (Sureshchandar & Leisten, 2006).

2.2.5 The Analytic Network Process (ANP).

The Analytic Network Process ANP is a multicriteria theory of measurement used to derive relative priority scales of absolutes numbers from individual judgments that also belong to

(34)

30 a fundamental scale of absolute numbers. These judgments represent the relative influence, of one or two elements over the other in a pair wise comparison process on a third element in the system, with respect to an underlying control criterion. Through its super matrix whose entries are themselves matrices of column priorities, the ANP synthesizes the outcome of dependence and feedback within and between clusters of elements. The Analytic Hierarchy Process (AHP) with its independence assumptions on upper levels from lower levels and the independence of elements in a lever is a special case of the ANP (Saaty, 2005).

The ANP provides a general framework to deal with decisions without making assumptions about the independence of higher-level elements from lower level elements and about the independence of elements within a level as in a hierarchy. In fact the ANP uses a network without the need to specify levels. As in the AHP dominance or relative importance is a central concept. In ANP, one provides a judgment from the fundamental scale of the AHP by answering two kinds of questions with regard to strength of dominance: 1) Given a criterion, which of two elements influences a third element more with respect to that criterion?, 2) Which of two elements influences a third element more with respect to a criterion? (Saaty, 2005).

The difference between a hierarchy and a network is shown on figure 4. A hierarchy has a goal or a source node or cluster. It also has a sink node or cluster known in probability theory as an absorbing state that represents the alternatives of the decision. It is a linear top down structure with no feedback from lower to higher levels. Unlike a hierarchy, a network spreads out in all directions and its clusters of elements are not arranged in a particular order. In addition, a network allows influence to be transmitted from a cluster to another one and back either directly from a second cluster or by transiting through intermediate clusters along a path which sometimes can return to the original cluster forming a cycle (Saaty, 2005).

(35)

31 Linear Hierarchy component, cluster (Level) element

A loop indicates that each element depends only on itself. Goal

Subcriteria Criteria

Alternatives

Feedback Network with Components having Inner and Outer Dependence among Their Elements

C₄ C₁

C₂ C₃

Feedback

Loop in a component indicates inner dependence of the elements in that component

with respect to a common property.

Arc from component C4to C2 indicates the outer dependence of the elements in C2on the elements in C4with respe to a common property.

Figure 5. How a Hierarchy Compares to a Network (Saaty, 2005).

Having been exposed to the AHP, the reader knows that criteria must be weighted. The weights cannot be meaningfully obtained by simply assigning numbers to them but need to be compared with an objective (or multiple objectives) in mind. . Comparisons require judgments. Judgments are associated with feelings, feelings with intensities, intensities with numbers, numbers with a fundamental scale, and a set of judgments represented by a fundamental scale to priorities. The fundamental scale that represents dominance of one element over another is an absolute scale and the priorities derived from it are normalized or idealized to again yield an absolute scale. Judgments are usually inconsistent. A modicum of inconsistency is a very useful fact because it indicates that our mind has the ability to learn new things that improve and even change our understanding. But large inconsistency can indicate lack of coherent understanding that may lead to a wrong decision (Saaty, 2005).

2.2.5.1 The ANP Mathematics.

Assume that we have a system of N clusters or components, whereby the elements in each component interact or have an impact on or are themselves influenced by some or all of the elements of that component or of another component with respect to a property governing the interactions of the entire system, such as energy or capital or political influence. Assume that component h, denoted by Ch, h = 1, ..., N, has nh elements, that we denote by

(36)

32 1, 2,..., nh

h h h

e e e . A priority vector derived from paired comparisons in the usual way represents the impact of a given set of elements in a component on another element in the system. When an element has no influence on another element, its influence priority is assigned (not derived) as zero (Saaty, 2005).

The priority vectors derived from pair wise comparison matrices are each entered as a part of some column of a supermatrix. The supermatrix shown in figure 3 represents the influence priority of an element on the left of the matrix on an element at the top of the matrix. (Saaty, 2005).

W_i1 W_i1 W_i1

Wij=

(j1) (j2) (jnj)

(j₁) (j₂) (jn_j)

W_i2 W_i2 W_i2

W_ini(j1) W_ini(j2) W_ini(jnj)

Figure 6. The Supermatrix of a Network and Detail of a Matrix in it. (Saaty, 2005).

The supermatrix must first be reduced to a matrix, each of whose columns sums to unity, resulting in what is known as a column stochastic matrix. A recommended approach by Saaty is to determine the relative importance of the clusters in the supermatrix with the column cluster as the controlling component. The resulting stochastic matrix is known as the weighted supermatrix (Saaty, 2005).

Raising a matrix to powers gives the long-term relative influences of the elements on each other. To achieve a convergence on the importance weights, the weighted supermatrix is raised to the power of 2k+1, where k is an arbitrarily large number, and this new matrix is called the limit supermatrix (Saaty, 2005) . The limit supermatrix has the same form as the weighted supermatrix, but all the columns of the limit supermatrix are the same. By normalizing each block of this supermatrix, the final priorities of all the elements in the matrix can be obtained.

C1 C2 CN

e11e12 e_1n1 e21e22 e_2n2 eN1eN2 e_NnN

W₁₁ W₁₂ W_1N

W₂₁ W₂₂ W_2N

W_N1 W_N2 W_NN

C1 C2 CN e11 e12 e_1n1 e21 e22 e_2n2 e_N1 eN2 e_NnN W=

C1 C2 CN

W₁₁ W₁₂ W_1N

W₂₁ W₂₂ W_2N

W_N1 W_N2 W_NN

C1 C2 CN e11 e12 e_1n1 e21 e22 e_2n2 e_N1 eN2 e_NnN

C1 C2 CN

W₁₁ W₁₂ W_1N

W₂₁ W₂₂ W_2N

W_N1 W_N2 W_NN

C1 C2 CN e11 e12 e_1n1 e21 e22 e_2n2 e_N1 eN2 e_NnN W=

(37)

33 ANP is argued to be able to solve all kinds of decision problems that we might encounter. It focuses on how to identify decision criteria (by structuring a decision model) and how to weight the criteria (by use of pair wise comparison). “The former is a powerful tool if the decision model is substantially affected by interdependent relationships” (Cheng, Heng, & Ling, 2005).

2.2.6 Summarizing AHP and ANP methodologies.

AHP and ANP are multicriteria decision-making tools, which are argued to possess qualitative (decision model development) and quantitative (decision model analysis) components. AHP models a hierarchical decision problem framework, which consists of multiple levels specifying unidirectional relationships. ANP models a network structure that relaxes the hierarchical and unidirectional assumptions in AHP to allow interdependent relationships in the decision making framework. Although the two decision tools possess the same qualitative and quantitative procedures to structure and analyze a decision problem, ANP needs further quantitative steps to solve a network decision problem (Cheng, Heng, & Ling, 2005). A brief description of the method is provided, which is based on Cheng, Heng, & Ling (2005) who suggested that ANP is composed of four qualitative (1 to 4) and five quantitative (5 to 9) steps:

1. To state the decision problem – The topmost level is to state the decision problem. This starts the decomposition of further levels down the structure until final level that is usually the scenarios or alternatives to be selected.

2. To make sure that the decision problem is to be solved by ANP or AHP – As already stated, ANP is used to structure a decision problem into a network form. For solving strictly hierarchical model, AHP is sufficient.

3. To structure the unstructured decision problem – The topmost decision problem level is abstract in nature. It must be decomposed into a set of manageable and measurable levels until the level of criteria for assessing the scenarios or alternatives.

4. To determine who the raters are – Those who are responsible for making the decision are the raters for completing a questionnaire.

(38)

34 5. To design a questionnaire for eliciting data from raters – It is suggested to use the pairwise comparison, which can elicit more information to assign weights to the rated elements. It is common to use the Fundamental Scale propose by Saaty shown in table 1.

6. To calculate the eigenvector of each of the developed matrices – Each decomposed level with respect to a higher level forms a matrix. It is necessary to calculate the eigenvector for the elements of this matrix.

7. To measure the consistency ratio (CR) of each of the matrices to find out the inconsistency of rating – One of the best reasons to use pairwise comparison and matrix is to measure the CR to ascertain that raters are consistent in rating. If the CR value cannot pass the acceptable level, it is certain that the raters rated arbitrarily or mistakenly. Rerating is then needed.

8. To form the supermatrix by the eigenvectors of the individual matrices. The eigenvectors of each of the developed matrices should gather together to form a supermatrix.

9. To compute the final limit matrix – In order to compute the final limit matrix, the supermatrix, which has been ensured of column stochastic, has to raise to high power until weights have been converged and remain stable.

2.2.7 Foresight.

The term comes from the Latin “prospicere” which means to look far. Thus, the forward looking concepts isolated envision the future and the possibilities that loom over him. What are the challenges we will face in the future? ¿Are we going to satisfy our expectations of life in the future?, those are questions that we frequently do. In this sense, foresight aims to stimulate the formulation of questions about things that concern us and affect us. Uncertainty about the future cannot be disclosed without first having the staff available to develop the opportunities and dangers for us tomorrow. "Foresight is ... seeing the future with optimism" (Sachs, 1980).

(39)

35 "We can try to take a pre-active stance in which we prepare for the expected changes, or go further and become proactive people who are working to achieve the desired future. It is in this attitude is where the spirit of strategic foresight” (Ozbekhan, 1968).

"The role of foresight is to provide the planner: a) a vision of desired future and b) a series of scenarios that define a wide range of options in terms of possible future" (Sachs, 1980).

2.2.8 Prediction, Projection and Forecasting.

A Chinese proverb says. “Prediction is difficult especially with regard to the future”. Prediction is one of the activities to which human beings dedicate much time and energy. We have incorporated in our daily lives a variety of predictive activities. We predict the weather, the daily lottery number, winning at sports, the value stocks, and even sex of the unborn child. However with all the sophistication that some techniques require we are still not able to make predictions accurately (Saaty & Vargas, 1991).

The past and the present are irreversible, unique and certain, but the knowledge we have of them is incomplete, the facts of the past form a small part of the unknown number of phenomena that make up reality. The history we know is one of many interpretations we have, even if the facts are unique (Saaty & Vargas, 1991).

The future is the domain of “becoming” and what if might be, It is not written, it is constructed or we drift into it. What happens in the future results only in part from our past actions, with their manifold and uncertain interpretations. The future evolves from the past, but there are many degrees of freedom, and hence, many possible futures (Saaty & Vargas, 1991).

We predict when we said in advance, foretell, or prophesy what is likely to happen in the future. We project when we calculate the numerical value associated with a future event. We forecast, a special kind of prediction, by relying on data of past happenings to generate or cast data for future happenings. Generally, one predicts (yes, no) a war, an earthquake or the outcome of a chess match, projects the value of the GNP or of unemployment, and forecasts the weather and, more scientifically, the economic trends. Prediction, projection and forecasting must be constrained in time and space: when and where. Often the accuracy

(40)

36 of a forecast is of interest along with how sensitive the outcome is to changes in the factors involved (Saaty & Vargas, 1991).

Prediction is the synthesis of past and present in an attempt to foretell the future. In general the purpose of prediction, whether of physical or social events, is to secure a measure of control over the potential outcome (Saaty & Vargas, 1991).

Our complex and interdependent world affects our lives in so many ways that we need new ways to anticipate the conditions of tomorrow. More and more, we all find ourselves speculating about predicting what will happen tomorrow, the day after, and even the next year. The businessman looks far ahead into the future to do his business today. Individuals everywhere try to be prepared for tomorrow, like corporations and governments. Because of the many variables that determine the future there is uncertainty in acting to control it (Saaty & Vargas, 1991).

In economics, prediction of cycles enables people and governments to plan their future with greater confidence, although the outcome of their plans may not always meet their expectations. Prediction is intimately tied to decision making because one of its goals is to provide guidance for taking action (Saaty & Vargas, 1991).

In all his decisions, the executive today needs to consider some kind of forecast or estimate of future customer needs and sales. Manufacturing, planning, and control are concern basically with the future. The past can no longer be controlled, the future can. It is necessary to estimate or guess what will happen from the present. The decision maker must factor potential future happenings in to his present decisions. If the manager is to cope with sudden changes in demand levels, price-cutting maneuvers by his competitors and by large swings of the economy. His decisions can be helped by good forecast of the future of his business (Saaty & Vargas, 1991).

2.2.8.1 Approaches to Prediction.

There are four general types of approaches to prediction (Saaty & Vargas, 1991): o Systematic generation of alternative paths to the future.

o Extrapolative trend examination.

(41)

37 o Historical analysis and analogy.

o Collective opinion techniques.

Systematic generation of alternatives paths to the future.

In this approach one attempt to construct some sort of systems model adequate to the kind of knowledge of the future that is desired and, from this, generate descriptions of alternative futures as functions of time. There are two types of models: quantitative and non-quantitative. Validation of the model is carried out by checking how well it describes the recent past (Saaty & Vargas, 1991).

Among the quantitative systematic methods the most frequently found in the literature are econometric models and combined econometric-demographic models.

Another well-known non-quantitative approach is “La prospective” This is a philosophy of looking at the future as an outcome of the desires and actions of actors influenced by an environment rather than continuation of action rather than a continuation of the past. La prospective, according to its first proponent, G. Berger, is neither a doctrine nor a system. It is a way of focusing and concentrating on the future by imaging it full blown and plausible, rather than by drawing deductions from the present.

Extrapolative trend examination

There are two major technical approaches to extrapolation (Saaty & Vargas, 1991): o Time series analysis.

o Causal models.

Time Series Analysis.

Extrapolative forecasting procedures are methods for modeling predictions of future observations based only on current and past observation. The literature contains a vast number of techniques ranging from the moving average, to the exponential moving average, to the more complete procedures based on models of Box and Jenkins.

(42)

38 Extrapolative approaches are familiar in economic and demographic analyses and in technology assessment.

Causal Models.

Regression models: This model relates scales to other variables and estimates an equation using the least-squares technique. Relationships are primarily analyzed statistically, although any relationships should be selected for testing on a rational ground.

Econometric models: An econometric model is a system of interdependent regressions equations. These models express causalities better than ordinary regression.

Intention-to-buy and anticipation surveys: These surveys of the general public determine intentions to buy certain products or derive an index that measures general feeling about the present and the future and estimates how this feeling will affect buying habits.

Input-output models: This model shows what flows of inputs must occur to obtain certain outputs.

Historical analysis and analogy (Saaty & Vargas, 1991).

Historical analyses can be useful, particularly in examining the hypothesis that industrialized society man currently be undergoing a profound transformation which will affect all aspects of the future.

Collective Opinion Techniques (Saaty & Vargas, 1991).

A fourth category of method used for future studies is that of gathering and processing individual judgments relating to the future. The type of methods encountered here range from surveys through Delphi.

Delphi Method.

A panel of experts is interrogated by a sequence of questionnaires in which the responses to one questionnaire are used to produce the next questionnaire. Any set of information

(43)

39 available to some experts and not others is passed on to others, enabling all the experts to have access to all the information for forecasting. This technique eliminates the bandwagon effect of majority opinion.

Market Research.

A systematic, formal, and conscious procedure for evolving and testing hypotheses about real markets.

Panel Consensus.

This technique is based on the assumption that several experts can arrive at a better forecast than one person. There is no secrecy, and communication is encouraged. The forecasts are sometimes influenced by social factors, and may not reflect true consensus.

Visionary Forecast.

A prophecy that uses personal insights, judgment, and when possible, facts about different scenarios of the future. It is characterized by subjective guesswork and imagination; in general, the methods used are nonscientific.

The Analytic Network Process combines these four approaches. It uses judgments from people to prioritize alternative futures and combine them into a single future of composite scenario. Scenarios, in general, may be of two types: exploratory (examine events that might influence the future and parametrize the principal components of the system) and anticipatory (conceptualize feasible and desired futures to discover alternatives and ways of actions to attain these futures). In turn, anticipatory scenarios can be divided into normative and contrast scenarios. The former determine the objectives that must be accomplished and also define a way for their realization. Contrast scenarios emphasize a set of assumptions which define the convex hull of possible futures (Saaty & Vargas, 1991).

I show in this thesis how to use the ANP approach to organize complexity subject to uncertainty and risk. We can apply our judgment that has been developed through experience, reason, and intuition, to assess and predict the likely future. We can work

(44)

40 forward from the present to the future, and we can also work backwards from the future by setting desired goals and finding out which policies we can adopt today to attain them.

2.2.9 Why the use of ANP for prediction?

Michel Godet in his book “From forecasting to “La prospective” shows some flaws on the classical forecasting methods, and here they are summarized (Godet, From Forecasting to "La Prospective": A New Way of Looking at Futures, 1982).

Forecasting models include a few explanatory variables, most of which can be easily quantified. They do not take into account the development of new relationships among variables and the possible changes in trends (Saaty & Vargas, 2006).

They usually assume that the dimension on which prediction take place is autonomous. For example, economic forecasting is divorced from social and political forecasting and is fragmented into technological, demographic, and other (Saaty & Vargas, 2006).

They forecast the future based solely on past data. For example, econometric models do not integrate qualitative and non-quantifiable parameters such as the behavior of relevant actors (Saaty & Vargas, 2006).

Classical models are both deterministic and structurally stable, which lead to error in forecasting, because of constant change which is selectively studied, and interpreted from a special point of view (Saaty & Vargas, 2006).

Conventional approaches of prediction and forecasting tend to be constrained by the estimated values of parameters and intercept terms. These are imbedded in the multi-equation models that are typically employed to produce "first-cut" forecasts of relevant endogenous variables. Additionally, the values of a large number of "exogenous" variables (relating to the future course of oil, the economic view, etc.) must be subjectively estimated on the basis of available evidence and consensus judgment. Initial forecasts produced by the raw models are then typically adjusted by "add" or "fudge" factors, most commonly in the form of shifts in the values of previously estimated intercept terms. This procedure is employed in order to produce forecasts that are consistent with recent values of key

(45)

41 endogenous variables when it is evident that a shift of some kind has occurred in portions of the underlying model structure. Such exercises also provide ample opportunity for resetting the values of exogenous variables (Saaty & Vargas, 2006).

This suggests that model builders are well aware of the limitations of their underlying models and the need to incorporate subjective judgments. However, these judgmental adjustments are necessarily non-systematic and ad hoc in nature. Accordingly, here we again utilize an alternative, the systematic approach ANP in order to remedy this deficiency (Saaty & Vargas, 2006).

(46)

42

Chapter 3 Model Development

3.1 Model Development.

In this chapter the model created to evaluate the different technologies and the timing in which the evaluations are going to take place is going to be explaining, in order to answer the questions which technology is going to replace the gasoline engine? and When is this replacement is going to take place?. The model was developed using software named Super Decisions, which is a specially made software for the ANP models.

[image:46.612.212.403.506.626.2]

The model use clusters where all the items related to this analysis are involved, first of all in order to answer the question which technology is going to replace the gasoline engine we evaluate the “Alternatives”. In this cluster we have the options taken under consideration to be the answer to this question (Hybrid, Hydrogen, Diesel, Gasoline and Ethanol) these options are going to be represented as you can see on figure 6.

Figure 7. Alternatives Cluster.