Multicriteria Decision

Making :Analytic
Hierarchy Process

Dr. Santosh Kumar

Faculty of Management Studies
University of Delhi
Delhi, India
Linear Programming Problems

Linear Programming Problems

•linear programming models were formulated and solved in order to optimize an objective
function value (a single measure of effectiveness) under a set of constraints.

•However, the optimization of such a single objective function is often not representative
of the reality due to divergent and conflicting objectives (economic as well as non-
economic) of any business, service or commercial organization.

•Consequently, there arises a need to attain a ‘satisfactory’ level of achievement amongst

multiple and conflicting objectives of an organization or a decision-maker.
Goal Programming
Goal Programming
Goal programming (GP) is a branch of Multi-Objective Optimization, which in
turn is a branch of Multi-Criteria Decision Analysis (MCDA). This is an optimization
program. It can be thought of as an extension or generalization of linear programming to
handle multiple, normally conflicting objective measures. Each of these measures is given
a goal or target value to be achieved. Unwanted deviations from this set of target values
are then minimized in an achievement function. This can be a vector or a weighted sum
dependent on the goal programming variant used. As satisfaction of the target is deemed
to satisfy the decision maker(s), an underlying satisficing philosophy is assumed.

Goal programming is used to perform three types of analysis:

•Determine the required resources to achieve a desired set of objectives.

•Determine the degree of attainment of the goals with the available resources.
•Providing the best satisfying solution under a varying amount of resources and priorities
of the goals.
Decision analysis

Decision analysis: is an analytic and systematic

approach to the study of decision-making. In this unit,
we will discuss the decision models that are useful in
helping managers make the best possible decisions.
 Multicriteria Decision
Making

•Decision analysis can be used to develop an optimal strategy when a

decision maker is faced with several decision alternatives and an uncertain
or risk-filled pattern of future events.

•Even when a careful decision analysis has been conducted, the uncertain
future events make the final consequence uncertain.

•Therisk associated with any decision alternative is a direct result of the

uncertainty associated with the final consequence.

•Good decision analysis includes risk analysis that provides probability

information about the favorable as well as the unfavorable consequences
that may occur.
 Multicriteria Decision
Making

 Decision-making is a common task in human beings’ daily lives. They often face
situations in which they need to analyze alternatives, which may be mutually exclusive,
and it is necessary to choose one of them.

To make the right decision about what alternative(s) could be the best or the most
suitable for the situation, empirical and scientific methods are used.

These decision situations may affect a wide range of problems, from very simple ones
such as choosing which shirt to wear to highly complex ones such as selecting the right
type of maintenance for a key tool in a complex engineering system, and so on.
 Multicriteria Decision
Making

therefore, depending on the characterization of the decision and

situation, different types of problems have been defined and structured
by decision theory to solve them.

The discipline of Multi-Criteria Decision-Making (MCDM) deals with

finding optimal results in complex scenarios in which scientific methods
are used by decision-makers to make decisions where more than one
criterion/attribute for evaluating each alternative is considered.

It has been used as a generic term for all techniques that assist
humans making decisions according to their preferences, in situations
where there are multiple conflicting criteria.

The approach of MCDM involves decision-making concerning

quantitative and qualitative factors.
 Multicriteria Decision
Making

Decisions can be made under different conditions depending on

distinct aspects such as the number of experts, the number of criteria,
or the definition environment.

Therefore, depending on the characterization of the decision and

situation, different types of problems have been defined and structured
by decision theory to solve them.
 MCDM: elements and
classification
1. It is not common even for decision problems with just one decision-
maker that she/he is clear about the evaluation of alternatives and
has only one single criterion. Most decision real-world problems are
usually evaluated by means of multiple and conflicting criteria.
2. Consequently, MCDM and the diversity of problems to which they are
successfully applied or may imply some heterogeneity in the
definitions of such problems.

Multiple criteria: each problem has multiple criteria, and depending on

the solution space of the MCDM problem, such criteria can be objectives
or attributes.
• Conflicting criteria: it is common that in the set of criteria, some of
them conflict with each other.
• Incommensurable units: often the criteria may involve different units
of measurement.
• Design/selection: the solutions of MCDM problems are either to design
the best alternative(s) or to select the best one among a predefined,
finite set of alternatives.
 MCDM: elements and
classification
It has been pointed out that criteria can be divided
into two types, objectives and attributes, which leads
to the classification of MCDM problems into two broad
categories :

1. Multi-objective decision-making (MODM) problems: these problems

are focused on continuous solution spaces and are solved by means of
mathematical programming models (linear programming, goal
programming, etc.) or metaheuristics.
2. Multi-attribute decision-making (MADM) problems: these are defined
on discrete solution spaces. For a better understanding and further
discussion of these types of MCDM problems, the basic concepts are
enumerated and briefly described as follows:
 MCDM: elements and
classification
•Criteria are the standard of judgment or rules to test acceptability. In MCDM-
specific literature, they may indicate objectives and/or attributes. Therefore,
an MCDM problem may mean either MODM or MADM; however, MCDM
usually means MADM.
•Objectives are the reflection of the decision-maker’s preference and indicate
the direction that she/he wishes to follow. MODM problems, as a result,
involve the design of alternatives that optimize or best satisfy the
decisionmaker’s objectives.
• Goals are elements desired by a decision-maker and are expressed in terms
of a specific state in space and time. Unlike objectives, which give the
desired direction, goals provide a desired target level to achieve.
• Attributes are the characteristics, qualities, or performance parameters
that define the alternatives of the MADM problem. The selection of the best
alternative implies the evaluation of the selected alternatives according to
the attributes.
• MODM problems are a continuous type of MCDM in which decision-makers
aim to achieve multiple objectives that are incommensurable and conflicting
with each other. An MODM model includes a vector of decision variables and
objective function(s) that describe the objectives, and constraints. Decision
makers attempt to optimize (maximize or minimize) the objective functions.
 Multi-criteria Decision
Making
BASIC DEFINITIONS:
Decision Maker: The decision maker(s) refers to the person(s), organization(s), or
stakeholder(s) to whom the decision problem under consideration belongs.
Decision Variable: A decision variable is defined as a factor over which the decision maker
has control. An example is a manufacturing company which has to decide how many of a
certain product to make in the next month. The set of decision variables fully describe the
problem and form the decision to be made. The purpose of the goal programming model can
be viewed as a search of all the possible combinations of decision variable values (known as
decision space) in order to determine the point that best satisfies the decision maker’s goals
and constraints.
Objective: An objective can be referred to as a criterion with the additional information of
the direction (maximize or minimize) in which the decision maker(s) prefer on the criterion
scale, for example, minimize cost or maximize the performance of a system. A decision
problem with a set of objectives to be maximized or minimized is referred to as a multi-
objective optimization problem. In practice, these objectives will be conflicting, that is they
cannot reach their optimal values simultaneously. If they could, then the model can be
solved as a single-objective problem for any of the objectives. The space formed by the
values of the set of objectives is known as objective space.
Aspiration Level: The numerical value specified by the decision maker that reflects his/her
desire or satisfactory level with regard to the objective function under consideration
 Multicriteria Decision
Making

•Goal: An objective function along with its aspiration level.

•Alternative: In MADM, there exists a number of predetermined,

limited, and independent alternatives, and each of them satisfies a level
of the desired attributes of the decision maker.

•Criterion: The criterion is the basis for evaluation, which means

measuring the effectiveness rate and is divided into the objective and
attribute. Objective: It is something pursued until its final achievement.
 Multicriteria Decision
Making
Thus A criterion is a single measure by which the goodness of any solution to a decision
problem can be measured. There are many possible criteria arising from different fields of
application but some of the most commonly arising relate at the highest level to:
• Cost
• Profit
• Time
• Distance
• Performance of a system
• Company or organizational strategy
• Personal preferences of the decision maker(s)
• Safety considerations
A decision problem which has more than one criterion is therefore referred to as a multi-
criteria decision making (MCDM) or multi-criteria decision aid (MCDA) problem. The
space formed by the set of criteria is known as criteria space.
 Multicriteria Decision
Making

Decision matrix: A matrix with the rank of which is generally

demonstrated as in the following Equation. where denotes alternatives
and , indicates attributes and represents the value of alternatives for
each attribute
 Multicriteria Decision
Making

Attribute: It is the property that should be in an alternative. Depending

on the idea of the decision maker, each alternative is associated with a
number of relevant attributes.
Positive attributes: They refer to the attributes of with positive
desirability from the perspective of decision maker; namely, their
greater amount is more favorable for the decision maker. Positive
attributes are usually as the profit,income or productivity.
Negative attributes: Attributes of with negative desirability from the
perspective of decision maker, which means that their lower amount is
more desirable to the decision maker. Negative attributes are usually as
loss or cost.
 Multicriteria Decision
Making
Non-compensatory attributes: The attributes in which the
disadvantage of an undesirable value in an attribute cannot be covered
by the advantage of a desirable value in another attribute.
Compensatory attributes: These attributes can interact with each
other; in other words, the disadvantage of an undesirable value in an
attribute can be covered by the advantage of a desirable value in
another attribute.
Independent attributes: Attributes which are absolutely uncorrelated
to the other attributes.
Dependent attributes: Attributes which are correlated to at least one
of the other attributes.
Quantitative attributes: They are attributes with a unit of
measurement, which are expressed numerically and are measurable.
Qualitative attributes: They are attributes usually without a unit of
measurement which can not be expressed numerically and are
immeasurable.
 Multicriteria Decision
Making

Three steps govern the process of MCDM analysis:

(1) Identification of the relevant criteria and

alternatives from the existing theory and practice.
(2) Assignment of numerical values to criteria to
indicate their relative importance and to quantify
the impacts of the alternatives on these criteria.
(3) Using a formal mathematical procedure for
analyzing numerical values to determine the ranking
(priorities) of the alternatives.
Single-criteria versus multi-criteria decision
making
Examples of MCDM in personal
life
Examples of MCDM in personal
life
Examples of MCDM in personal
life
Examples of MCDM in
professional life
Examples of MCDM in
professional life
Examples of MCDM in
professional life
Examples of MCDM in
professional life
 Overview: AHP
The Analytic Hierarchy Process (AHP) is a structured technique for
organizing and analyzing complex decisions, based on Mathematics and
Psychology.
The analytical hierarchy process (AHP) is a method used for complex decision-making in which
various subjective components are quantified and prioritized as per the decision-maker’s own
criteria. The final priority order can be used objectively to arrive at the decision.

 It was developed by Thomas L. Saaty in the 1970s and has been

extensively studied and refined since then.

It has particular application in group decision making.

AHP has been applied in variety of decisions and planning projects in

nearly 20 countries.

Rather than prescribing a "correct" decision, the AHP helps decision

makers find one that best suits their goal and their understanding of
the problem. It provides a comprehensive and rational framework for
structuring a decision problem, for representing and quantifying its
elements, for relating those elements to overall goals, and for
evaluating alternative solutions.
 Overview: AHP

 When a decision is dependent on multiple criteria and/or involves subjectivity, AHP

framework can be used to arrive at an optimal decision in a rational and structured
fashion.
 To reduce bias in decision making, AHP reduces complex decisions involving several
criteria and alternatives into pairwise comparisons and provides a quantitative
methodology that ensures consistency in the decision-maker’s evaluations.
 Key elements of decision making using AHP include an objective (goal), criteria,
sub-criteria, and alternatives (final options to choose from).
 A hierarchical structure of the elements is created with a goal at the top, required
criteria and sub-criteria in subsequent layers, and alternatives as the last layer.
 Simple AHP models might not have sub- criteria and the hierarchy may contain only
three layers.
 Five Steps of AHP

1. Define objective (goal) and gather elements of decision

making such as criteria, sub-criteria, and alternatives.
2. Develop a hierarchical structure of the elements.
3. Make a pairwise comparison of elements in each group,
to determine the relative importance of different criteria
with respect to the goal.
4. Calculate weighting and consistency ratio.

5. Evaluate alternatives according to weighting.

AHP-STRUCTURE
AHP-PROCESS
Typical application areas
Resource Apportioning resources among a set of alternatives
allocation
Total Quality Dealing with the multidimensional aspects of quality
Management and quality improvement
Hiring, evaluating Current/ Prospective employees can be ranked by
and promoting considering all required criteria. The ranking can
employees assist the organization to select the best employee.
Strategic planning In deriving strategies people usually try to achieve
multiple, and sometimes conflicting objectives.
Moreover, specific strategies are normally
determined by a board of directors whose members
have different backgrounds, judgments and personal
goals. Thus, AHP comes into play.
Facility Location The AHP model provides a framework to assist
managers in analyzing various location factors,
evaluating location site alternatives, and making
final location selections. The primary principle of the
AHP model is to match decision‐makers’ preferences
with location site characteristics.
Vendor Supplier selection is defined in as the ‘‘process of finding
selection the suppliers being able to provide the buyer with the
right quality products and/or services at the right price,
at the right quantities and at the right time”. To choose
the right supplier, different methods can be used, one of
which is AHP.
City Rank a set of cities from the most to the least livable. The
Evaluation satisfaction of the final goal ‘livability’ can be
and Planning decomposed into the satisfaction of some criteria, such as
‘environment’, ‘services’, security’, and each of these
criteria can itself be decomposed into sub criteria.
Country The alternatives are the countries themselves, and the
Ranking criteria simply all those characteristics which could make
one country better than another.
Mobile Value AHP can be used to determine the most important
Services decision criteria driving the customers’ adoption of
mobile devices and mobile services. The main attributes
that can be considered include payment mode,
functionality, added value, perceived quality, cost, and
performance. The results of this type of studies can be
essential for various service providers (operators, mobile
handset manufacturers) to design profitable applications
that generate value for the end-users.
Organ Lin and Harris proposed to use the AHP to decompose the
Transplant four criteria ‘urgency’, ‘efficiency’, ‘benefit’, and ‘equity’ into
sub criteria and eventually estimate their importance in the
donors-organs matching process. Patients were treated as
alternatives, but it is clear that their huge amount would
have made the use of subjective judgments impossible.
Fortunately, in this case, the pairwise comparison matrices at
the alternative level were filled automatically since different
criteria were quite easily quantifiable.
Chess The AHP has been used for forecasting too. In sports,
Prediction athletes can be seen as alternatives and their characteristics
as criteria, and the player rated the highest shall be regarded
as the most likely to win. Here we refer to an application of
the AHP for the prediction of winners in chess matches—The
AHP was used to evaluate the outcome of the Chess World
Championships as well as of the matches between Fischer
and Spassky in 1972 and Karpov and Korchnoi in 1978.
Evaluating An acquiring unit normally does not stop after one
mergers acquisition and has to formulate an overall mergers and
and acquisitions strategy. Moreover, every single decision within
acquisitio this strategy could have a major impact on its future. Hence,
ns this topic is perfectly suited for the Analytic Hierarchy
Process (AHP), especially when we consider the amount of
uncertainty and the many conflicts involved in such
 Fundamentals

The AHP can be applied to a multitude of decision making

problems involving a finite number of alternatives. Formally, in
this setting, in a decision process there is one goal and a finite set
of alternatives,
X = {x1, . . . ,xn}, from which the decision maker, is usually asked
to select the best one.
 Fundamentals
Often, in decision processes, the decision maker is asked to assign
a score to each alternative and then to choose the one with the
maximum value.

That is, given a set of alternatives, X = {x1, . . . ,xn}, the decision

maker should provide a weight vector w = (w1, . . . ,wn)T , where wi
is a value which coherently estimates the score of alternative xi.

That is, the greater wi, the better the ith alternative. Similarly to
what happens for value theory the rule is that alternative xi is
preferred to alternative xj if and only if wi > wj .

Weight vectors are nothing else but ratings, and their components
wi are called priorities, or weights, of the alternatives xi. For
example,

w = (0.4,0.2,0.3,0.1)T implies x1 f x3 f x2 f x4 where xi f xj means

A decision maker could run into troubles when asked to submit a
rating in the form of a numerical vector for a large number of
alternatives.

Does not it often happen that we cannot decide among several

alternatives? Even worse, do not we decide and eventually realize
that it was not the best decision?

This is a matter of fact and originates from our cognitive limits

and the impossibility of effectively comparing several alternatives
at the same time.

The strategy is that of decomposing the original problem into

many smaller sub-problems and deal with these latter ones.
Formally, the pairwise comparisons are collected into a pairwise
comparison matrix, A = (aij)n×n, structured as follows:

A= a11 a12 . . . a1n

a21 a22 . . . a2n
...
...
...
...
an1 an2 . . . ann
(1)

with aij > 0 expressing the degree of preference of xi to xj .

More precisely, according to Saaty’s theory, each entry is
supposed to approximate the ratio between two weights

aij ≈ wi ∀ i, j.
(2)
wj

This means that, if the entries exactly represent ratios between

weights, then the matrix A can be expressed in the following form,

A = (wi / wj)n×n = w1 /w1 w1 /w2 . . . w1 /wn

w2 /w1 w2 /w2 . . . w2 /wn
...
...
...
...
wn /w1 wn /w2 . . . wn /wn
(3)
Note that, as soon as we account for (2) and consider (3), a
condition of multiplicative reciprocity aij = 1/aji ∀ i, j holds, and A
can be simplified and rewritten,

A= 1 a12 ・・・ a1n

1 1 ・・・ a2n
a12
...
...
...
...
1 1 ・・・ 1
a1n a2n
(4)

In words, the simplified structure of pairwise comparison matrices

in this form follows from the assumption that if, for example, x1 is
2 times better than x2 , then we can deduce that x2 is 1/2 as good
as x1.
 Methods to obtain the Priority
Vector

A single decision maker is perfectly rational and can precisely express his preferences on all
pairs of independent alternatives and criteria using positive real numbers.

1. Eigenvector Method

2. Geometric Mean Method

3. Other Methods
 Eigenvector Method
The most popular method to estimate a priority vector is that proposed by
Saaty himself, according to which the priority vector should be the
principal eigenvector of A. In linear algebra it is often called the Perron-
Frobenius eigenvector, from the homonymic theorem. The method stems
from the following observation. Taking a matrix A whose entries are
exactly obtained as ratios between weights and multiplying it by w one
obtains

Aw = w1/w1 w1/w2 ... w1/wn w1 = nw1 =

nw.

w2/w1 w2/w2 ... w2/wn . .

... .
.
... .
.
... wn
nwn
...
w /w w /w ... w /w
Moreover by knowing that the other eigenvalue of A is 0, and has
multiplicity (n−1), then we know that n is the largest eigenvalue of A.
Hence, if the entries of A are ratios between weights, then the weight
vector is the eigenvector of A associated with the eigenvalue n. Saaty
proposed to extend this result to all pairwise comparison matrices by
replacing n with the more generic maximum eigenvalue of A. That is,
vector w can be obtained from any pairwise comparison matrix A as the
solution of the following equation system,

Aw =λmaxw
wT 1 = 1

where λmax is the maximum eigenvalue of A, and 1 = (1, . . . ,1)T . Although

this problem can easily be solved by mathematical software and also
spreadsheets, its interpretation remains cumbersome for practitioners.
 Geometric Mean Method
Another widely used method to estimate the priority vector is the
geometric mean
method, proposed by Crawford and Williams. According to this method
each
component of w is obtained as the geometric mean of the elements on the
respective
row divided by a normalization term so that the components of w
eventually add up
to 1,
It is apparent that the geometric mean method is very appealing or
practical applications since, in contrast to the eigenvector method, the
weights can be expressed as analytic functions of the entries of the matrix.
Furthermore, even the final weights of the whole hierarchy can be
expressed as analytic expressions of the entries of all the matrices in the
hierarchy. This is particularly important since it opens avenues to perform
efficiently some sensitivity analysis. Moreover, on a more mathematical
note, it is interesting to note that the vector w obtained with this method,
can equivalently be obtained as the argument minimizing the following
optimization problem:
 Other Methods
A large number of alternative methods to compute the priority vector have been
proposed in the literature. Choo and Wedley listed 18 different methods and
proposed a numerical and comparative study. Lin reconsidered and simplified their
framework. Another comparative study was offered by Ishizaka and Lusti. Instead,
Cook and Kress presented a more axiomatic analysis where some desirable
properties were stated. From all these studies it appears that, besides the
eigenvector and the geometric mean method, other two methods have gained some
popularity.
• The so-called least squares method where the priority vector is the argument
solving
the following optimization problem:

In spite of its elegance, this optimization problem can have local minimizers
where the
optimization algorithms get trapped.
• The other one is the normalized columns method which requires the normalization
of all the
 Consistency
A perfectly rational decision maker should be able to state his pairwise
preferences exactly, i.e. aij = wi/wj ∀ i, j. So, let us consider the
ramifications of this condition on the entries of the pairwise comparison
matrix A. If we write aijajk and apply the condition aij = wi/wj ∀ i, j, then we
can derive the following

aij ajk = wi wj = wi = aik.

wj wk
wk

Hence, we discovered that, if all the entries of the pairwise comparison

matrix A satisfy the condition aij = wi /wj ∀i, j, then the following condition
holds,

aik = aijajk ∀ i, j, k ,

which means that each direct comparison aik is exactly confirmed by all
indirect comparisons aijajk ∀ j. Formally, a decision maker able to give
 An Illustration of AHP
Car A Car D
Car B Car C

Which of the four cars should I

purchase?
 Step 1: Structure a
Hierarchy

Tasks Example
Define the problem Which of the four cars to purchase?

Style Visual appeal, Aesthetics

Reliability Strength, Robustness
Determine the criteria
Fuel Mileage, Maintenance
Economy
Cost Extent to which it fits the budget
Identify the alternatives Car A, Car B, Car C, Car D
 Decomposition of the Problem into a
Hierarchy

Overall
Goal

Criteria

Alternativ
es
 Step 2: Make Pairwise
Comparisons
Arrange the elements in the second level into a matrix and elicit
judgments from the people who have the problem about the relative
importance of the elements with respect to the overall goal,
satisfaction with the car. The scale to use in making the judgments is
given on the next page. This scale has been validated for
effectiveness, not only in many applications by a number of people,
but also through theoretical comparisons with a large number of
other scales.
 Saaty’s Verbal Scale

Intensity of
importance on an Definition Explanation
absolute scale
1 Equal importance Two activities contribute equally to the objective
3 Moderate importance of Experience and judgment strongly favor one activity
one over another over another
5 Essential or strong Experience and judgement strongly favor one activity
importance over another
7 Very strong importance An activity is strongly favored and its dominance
demonstrated in practice
9 Extreme importance The evidence favoring one activity over another is of
tile highest possible order of affirmation
2,4,6,8 Intermediate values When compromise is needed
between the two adjacent
judgments

Reciprocals If activity i has one of the above numbers assigned to it when compared with activity
j, then j has the reciprocal value when compared with i
Rationals Ratios arising from the scale If consistency were to be forced by obtaining n
numerical values to span the matrix
 Pairwise Comparison Matrix
(Theory)

 Element Ci,j of the matrix is the measure of preference of the item

in row i when compared to the item in column j.

 AHP assigns ‘1’ to all elements on the diagonal of the pairwise

comparison matrix.
When we compare any alternative against itself (on the criterion) the
judgment must be that they are equally preferred.

 AHP obtains the preference rating of Cj,i by computing the

reciprocal (inverse) of Ci,j (the transpose position).
The preference value of 2 is interpreted as indicating that alternative i is
twice as preferable as alternative j. Thus, it follows that alternative j
must
be one-half as preferable as alternative i. This is called Reciprocacity.

 Transitivity holds. If one has to select one colour and he very strongly
prefers blue to green, and strongly prefers blue to yellow, then given a
choice between green and yellow, he should strongly prefer yellow to
green.

According to the above rules, the number of entries actually filled in

by decision makers is (n2 – n)/2, where n is the number of elements to
be compared.
 Pairwise Comparison Matrix
(Example)

Criterion Style Reliability Fuel Economy

Style 1 1/2 3

Reliability 2 1 4

Fuel Economy 1/3 1/4 1

 Procedure for Synthesizing
Judgements

1. Eigenvector Method

The following three-step procedure provides a good

approximation of the synthesized priorities:

Step 1: Sum the values in each column of the pairwise

comparison
matrix.

Step 2: Divide each element in the pairwise matrix by its

column total.
The resulting matrix is referred to as the
normalized pairwise
comparison matrix.

Step 3: Compute the average of the elements in each row of

the
 2. Row – Geometric Mean Method (RGM)

Take the nth root of the product of elements of each row of

the original pairwise comparison matrices (i.e., conversion
to decimals or normalization not required).

Here we use the post normalization method, i.e., we

normalize after applying RGM. That is called NRGM.
Eigenvector
Method Step 1: Sum the values in each
column

Style A B C D
A 1 5 2 3
B 1/5 1 1/3 1/3
C 1/2 3 1 3
D 1/3 3 1/3 1

Style A B C D
A 1.00 5.00 2.00 3.00
B 0.20 1.00 0.33 0.33
C 0.50 3.00 1.00 3.00
D 0.33 3.00 0.33 1.00
Column
2.03 12.00 3.67 7.33
Total
Step 2: Divide each element in the pairwise matrix by its
column total

Style A B C D

A 0.49 0.42 0.55 0.41

B 0.10 0.08 0.09 0.05

C 0.25 0.25 0.27 0.41

D 0.16 0.25 0.09 0.14

Step 3: Compute the average of the elements in each row of the
normalized matrix

Row Priori
Style A B C D
Average ty
A 0.49 0.42 0.55 0.41 0.47 Vecto
 0.47
r 
B 0.10 0.08 0.09 0.05 0.08  0.08 
 
C 0.25 0.25 0.27 0.41 0.29  0.29 
 
D 0.16 0.25 0.09 0.14 0.16  0.16 
 Consistency

 An important consideration in terms of the quality of the ultimate

decision relates to the consistency of judgments that the decision maker
demonstrates during the series of pairwise comparisons.
It should be realized that perfect consistency is very difficult to achieve and
that some lack of consistency is expected to exist in almost any set of pairwise
comparisons.

 To handle the consistency question, the AHP provides a method for

measuring the degree of consistency among the pairwise judgments
provided by the decision maker.
• If the degree of consistency is acceptable, the decision process can
continue.
• If the degree of consistency is unacceptable, the decision maker
should
reconsider and possibly revise the pairwise comparison judgments
before
proceeding with the analysis.

 The AHP provides a measure of the consistency of pairwise comparison

judgments by computing a CONSISTENCY RATIO
• The ratio is designed in such a way that values of the ratio
exceeding 0.10 are indicative of inconsistent judgments.
 Estimating Consistency Ratio
(Theory)

Step 1: Multiply each value in the first column of the

pairwise comparison matrix by the relative priority of the
first item considered. Perform same procedure for the
other items. Sum the values across the rows to obtain a
vector of values labeled “weighted sum.”

Step 2: Divide the elements of the vector of weighted sums

obtained in Step 1 by the corresponding priority value.

Step 3: Compute the average of the values computed in Step 2.

This average is denoted as λmax.
λ max  n
CI 
Step 4: Compute the consistency index (CI): n 1
Where n is the number of items being compared
CI
CR
Step 5: Compute the consistency ratio (CR): RI 

Where RI is the RANDOM INDEX

 Random Index

Random index (RI) is the consistency index of a randomly

generated pairwise
comparison matrix.
RI depends on the number of elements being compared (i.e.,
size of pairwise
comparison matrix) and takes on the following values:
n 1 2 3 4 5 6 7 8 9 10

RI 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49
 Estimating Consistency Ratio
(Example)
Step 1: Multiply each value in the first column of the
pairwise comparison matrix by the relative priority of the
first item considered. Perform same procedure for the
other items. Sum the values across the rows to obtain a
vector of values labeled “weighted sum.”

Priori
Style A B C D
ty
Vecto
A 1 5 2 3  0.47
r 1.93 
B 1/5 1 1/3 1/3
 0.08  0.32
   
C 1/2 3 1 3  0.29 1.25 
   
D 1/3 3 1/3 1
 0.16  0.65
 Estimating Consistency Ratio
(Example)

Step 2: Divide the elements of Step 3: Compute the average

the vector of weighted of the values
sums obtained in Step computed in Step 2.
1 by the This average is
corresponding priority denoted as λmax
value
1.93/0.47   4.15  4.15  4.08  4.23  4.07
 0.32/0.08   4.08  λ max  4.13
    4
1.25/0.29   4.23 
   
 0.65/0.16   4.07 
 Estimating Consistency Ratio
(Example)

Step 4: Compute the Step 5: Compute the

consistency index (CI) consistency ratio
(CR)

λ max  n 4.13  4 CI 0.04

CI   0.04 CR   0.05
n 1 4 1 RI 0.90
The degree of consistency exhibited
in the pairwise comparison matrix for
comfort is acceptable.
 Comparison Matrices and Local
Priorities
Row
Reliabilit
y A B C D Averag λmax CI RI
e
A 1 3 1/3 2 0.23

B 1/3 1 1/5 1/2 0.08

4.05 0.02 0.02
C 3 5 1 4 0.54

D 1/2 2 1/4 1 0.14

Row
Fuel
Economy A B C D Averag λmax CI RI
e
A 1 1/7 1/3 1/5 0.06
B 7 1 4 3 0.55
4.07 0.02 0.03
C 3 1/4 1 1/2 0.14
D 5 1/3 2 1 0.25
 Priorities of the Criteria

Criterion Style Reliability Fuel Economy Priorities

Style 1 1/2 3 0.32

Reliability 2 1 4 0.56

Fuel Economy 1/3 1/4 1 0.12

 Step 3: Determine the Best
Alternative
To obtain the final priorities, sum the product of the criterion
priority (with respect to the overall goal) times the priority of
the decision alternative with respect to that criterion.

Fuel Criteri Global

Reliabili
Style Econom a Prioriti
ty
y Prioriti es
A 0.47 0.23 0.06
es  0.29
 0.32  0.14
B 0.08 0.08 0.55  0.56  
   0.41
C 0.29 0.54 0.14
 0.12  
D 0.16 0.14 0.25  0.16
 Car C
Car C
Car A

Car D

Car B

C f A f D f B i.e., Car C is the most preferred

alternative
 Advantages of AHP
1. It allows Multi-Criteria Decision Making.

2. It is applicable when it is difficult to formulate criteria

evaluations, i.e., it allows qualitative evaluation as well as
quantitative evaluation.

3. It is applicable for group decision making environments

where the group members can use their experience, values
and knowledge to break down a problem into a hierarchy.
The members of the group may vote for the criteria and
sub-criteria to form the AHP tree and the overall priorities
obtained from members of the group.
 Disadvantages of AHP
1. There are hidden assumptions like consistency check which may
require repeating evaluations, making the process cumbersome.

2. Difficult to use manually when the number of criteria or sub-

criteria is high.

3. It is difficult to add a new criteria or sub-criteria or to take out

an existing criteria or sub-criteria since there could be drastic
changes in the overall ranking. This is called the Rank –
Reversal Phenomenon i.e.,
if a new alternative is added to the original set of alternatives,
then the order relation on the old set of alternatives should not
change. Transposing this concept to our daily lives, if one has to
select one Colours
colour and he very strongly prefers blue to green, and
strongly prefers blue to yellow, then given a choice between
green and yellow, he should1 strongly9prefer yellow5 to green.
1/9 1 1/4
1/5 3 1
4. Different Scales
In everyday life, people are more inclined to use linguistic
expressions like “I
slightly prefer pasta with salmon to pasta with cheese” or “I
strongly prefer one
banana to one apple”. To help the decision maker, some
linguistic expressions
have been proposed and then linked to different values
assignable to the
entries aij. Hence, the decision maker can express opinions on
pairs using
linguistic terms, which are then associated to real numbers.
In his original paper on the AHP, Saaty proposed an association
between verbal
judgments and values for pairwise comparisons. Other scales
have been
proposed and studied, among others, by Ji and Jiang to which
the reader can
refer for a short overview. One of the foremost is the balanced
scale proposed by
P¨ohj¨onen et al. The balanced scale, Saaty’s scale, and their
matching with
verbal judgments are reported in the following table:
 Examples

1.Accepting a Job Offer

2.Problem of Selecting a House

 Example 1

A employee is selecting a new job to expand his/her domain

knowledge . The employee want to use AHP to help him/her
decide which job offer to take up. The employee has four
criteria on which he/she wants to base the decision:

1. Location

2. Salary

3. Job Content: Since the focus is on expanding domain

knowledge, this is
further classified into:
* Creativity
* Tools Used

4. Long Term Prospects

 Accepting a Job
Offer

Job Prospec
Location Salary
Content ts

Creativit Tools
y Used
 RGM
Method Pairwise Comparison Matrix
(Criteria)
Job
Criterion Location Salary Prospects
Content
Location 1 1/7 1/5 1/9
Salary 7 1 2 1/2
Job
5 1/2 1 1/3
Content
Prospects 9 2 3 1
Criterio Locatio Job Prospec
Salary RGM NRGM
n n Content ts
Locatio
1 1/7 1/5 1/9 0.24 0.04
n
Salary 7 1 2 1/2 1.63 0.29
Job
5 1/2 1 1/3 0.96 0.17
Content
Prospec
9 2 3 1 2.71 0.49
ts
Creativi
A B C D RGM NRGM
ty
A 1 5 1/2 3 1.65 0.30
B 1/5 1 1/9 1/7 0.24 0.04
C 2 9 1 3 2.71 0.49
D 1/3 7 1/3 1 0.94 0.17
Column Total 5.54 1.00
λmax = 4.21 CI = 0.07 CR = 0.08
Tools A B C D RGM NRGM
A 1 1/7 1/5 1/9 2.34 0.48
B 7 1 2 1/2 1.32 0.27
C 5 1/2 1 1/3 0.76 0.16
D 9 2 3 1 0.43 0.09
Column Total 4.84 1.00
λmax = 4.01 CI = 0.00 CR = 0.01
 Job Content Sub – Criteria Weights

Criterio Creativi
Tools RGM NRGM
n ty
Creativi
1 1/7 2.65 0.875
ty
Tools 7 1 0.38 0.125
Column Total 3.02 1.00

Therefore, Job Content Priority

Vector is:
Job
Creativi Sub-
Tools Content
ty Criteria
Priorities
Priorities  0.32 
A 0.30 0.48
B 0.04 0.27  0.875  0.07 
 0.125  
C 0.49 0.16    0.45 
D 0.17 0.09  
 0.16 
Locatio
A B C D RGM NRGM
n
A 1 1/7 2 1/3 0.56 0.09
B 7 1 8 3 3.60 0.60
C 1/2 1/8 1 1/5 0.33 0.06
D 3 1/3 5 1 1.50 0.25
Column Total 5.99 1.00
λmax = 4.05 CI = 0.02 CR = 0.02
Prospec
A B C D RGM NRGM
ts
A 1 5 1/4 2 1.26 0.21
B 1/5 1 1/9 1/7 0.24 0.04
C 4 9 1 5 3.66 0.60
D 1/2 7 1/5 1 0.91 0.15
Column Total 6.07 1.00
λmax = 4.24 CI = 0.08 CR = 0.09
 Salary Weights

Since Salary is a Quantitative Variable, we

simply normalize the values to get the
priority vector

Salary Normalize
(in Lacs per Annum) d values
A 8.5 0.24
B 9 0.25
C 7.8 0.22
D 10.2 0.29
Column
35.5 1.00
Total
 Global Priorities

Locatio Job Prospec Criteri Global

Salary
n Content ts a Prioriti
Prioriti es
A 0.09 0.24 0.32 0.21  0.04
es  0.23
B 0.60 0.25 0.07 0.04
 0.29  0.13
   
C 0.06 0.22 0.45 0.60  0.17  0.44
   
D 0.25 0.29 0.16 0.15
 0.49  0.20

C f A f D f B i.e., Job C is the most preferred

alternative
 Example 2

When advising a family of average income to buy a house, the

family identified eight criteria which they thought they had to
look for in a house. These criteria fall into three categories:

1. Economic

2. Geographic

3. Physical

Although one may have begun by examining the relative

importance of these clusters, the family felt they wanted to
prioritize the relative importance of all the criteria without
working with clusters. The problem was to decide which of
three candidate houses to choose.
Note: In this example, Consistency Ratios marked with arrows
do not
conform to the requirement of being ≤ 0.10 (10%).
Hence, those matrices
are inconsistent and need to be revised. The final
priorities have been
House A
This house is the largest of them all.
It is located in a neighborhood with
little traffic and low taxes. Its yard
space is comparably larger than
houses B and C. However, the
general condition is not very good
and it needs cleaning and painting.
Also, the financing is unsatisfactory
because it would have to be bank-
financed at high interest.
House B This house is a little smaller than
House A and is not close to a bus
route. The neighborhood gives one
the feeling of insecurity because of
traffic conditions. The yard space is
fairly small and the house lacks the
basic modern facilities. On the other
hand, the general condition is very
good. Also, an assumable mortgage is
obtainable which means the financing
is good with a rather low interest
rate.
House C
House C is very small and has few
modern facilities. The neighborhood
has high taxes, but is in good
condition and seems secure. The yard
space is bigger than that of House B,
but is not comparable to House A's
spacious surroundings. The general
condition of the house is good and it
has a pretty carpet and drapes.
 Tasks Example
Define the problem Which of three candidate houses to choose?
(1) Size of house: Storage space; size of rooms,
number of
rooms; total area of house
(2) Location to Convenient, close bus
service
bus lines:
(3) Neighborhood: Little traffic, secure, nice
view, low taxes,
good condition of
neighborhood
(4) Age of house: Self-explanatory
(5) Yard space: Includes front, back and side,
and space
Determine the
from neighbors
criteria
(6) Modern Dishwashers, garbage disposals, air

facilities: conditioning, alarm system, and

other
such items possessed by a
house
(7) General Repairs needed, walls, carpet,
drapes,
 Satisfaction with
House

Yard Modern General

Size Transportation Neighbourhood Age Financing
Space Facilities Condition

House House House

A B C
Pairwise Comparison Matrix
(Criteria)
Comparison Matrices and Local
Priorities
Comparison Matrices and Local
Priorities
Comparison Matrices and Local
Priorities
Comparison Matrices and Local
Priorities
House A

A f B f C i.e., House A
is the most preferred
alternative
CASE STUDY 1:(RP)

Prioritizing Humanitarian Supply Chain

Challenges
(AHP-APPROACH)

Please see the separate document

CASE STUDY 2:
(Harved/Ivey)

Please see the separate document

CASE STUDY 3:
(Harved/Ivey)

Please see the separate document

Thank
You!