Analytic Hierarchy Process

 Humans are not often logical creatures.

 Most of the time we base our judgments on hazy
impressions of reality and then use logic to
defend our conclusions.
 The AHP organizes feelings, intuition, and logic
in a structured approach to decision making.
Introduction
 The analytic hierarchy process (AHP) was developed
by Thomas L. Saaty.
 Saaty, T.L., The Analytic Hierarchy Process, New York:
McGraw-Hill, 1980
 The AHP is designed to solve complex problems
involving multiple criteria.
 An advantage of the AHP is that it is designed to
handle situations in which the subjective judgments of
individuals constitute an important part of the
decision process.
Introduction
 Basically the AHP is a method of
(1) Breaking down a complex, unstructured situation into
its component;
(2) Arranging these parts, or variables into a hierarchic
order;
(3) Assigning numerical values to subjective judgments
on the relative importance of each variable; and
(4) Synthesizing the judgments to determine which
variables have the highest priority and should be acted
upon to influence the outcome of the situation.
Introduction
 The process requires the decision maker to
provide judgments about the relative
importance of each criterion and then specify
a preference for each decision alternative on
each criterion.
 The output of the AHP is a prioritized ranking
indicating the overall preference for each of
the decision alternatives.
Major Steps of AHP
1) Develop a graphical representation of the problem in terms of
the overall goal, the criteria, and the decision alternatives.
(i.e., the hierarchy of the problem)
2) Specify his/her judgments about the relative importance of
each criterion in terms of its contribution to the achievement
of the overall goal.
3) Indicate a preference or priority for each decision alternative
in terms of how it contributes to each criterion.
4) Given the information on relative importance and preferences,
a mathematical process is used to synthesize the information
(including consistency checking) and provide a priority
ranking of all alternatives in terms of their overall preference.
Hierarchy Development
 The first step in the AHP is to develop a graphical
representation of the problem in terms of the overall
goal, the criteria, and the decision alternatives.

Overall Goal: Select the Best Car

Criteria: Price MPG Comfort Style

Decision Car A Car A Car A Car A

Alternatives:
Car B Car B Car B Car B
Car C Car C Car C Car C
Pairwise Comparisons
 Pairwise comparisons are fundamental building
blocks of the AHP.
 The AHP employs an underlying scale with
values from 1 to 9 to rate the relative
preferences for two items.
Pairwise Comparison Matrix
 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 a 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.
 According above rules, the number of entries actually filled in
by decision makers is n(n – 1)/2, where n is the number of
elements to be compared.
Preference Scale
Verbal Judgment of Preference Numerical
Rating
Extremely preferred 9
Very strongly to extremely preferred 8
Very strongly preferred 7
Strongly to very strongly preferred 6
Strongly preferred 5
Moderately to strongly preferred 4
Moderately preferred 3
Equally to moderately preferred 2
Equally preferred 1
Synthesis
 The procedure to estimate the relative priority for
each decision alternative in terms of the criterion is
referred to as synthetization.
 Once the matrix of pairwise comparisons has been
developed, priority of each of the elements (priority of each
alternative on specific criterion; priority of each criterion
on overall goal) being compared can be calculated.
Procedure for Synthesizing Judgments
 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.  a a a 
11 12 1n
 n n
... n 
  ai1 a a in 
 a11 a12 ... am1  i2
 i 1 i 1 i 1 
 a a2 n 
a a22 ... am 2   n 21
a22
... 
A   21
n n

, A'  

ai1 a i2  ain  ,

 ... ... ... ...  
i 1 i 1 i 1

 ... ... ... ... 
   a ann 
 an1 an 2 ... ann  an 2
 n n1 n
... n 
 ain 
 
 i 1
ai1 a
i 1
i2 
i 1


Step 3: Compute the average of the elements in each row of
the normalized matrix.
 These averages provide an estimate of the relative priorities
of the elements being compared.
 a11 a12 a1n 
 n  n
 n 
  ai1  ai 2  ain 
 i 1 i 1 i 1 
 n 
 a a a   c1 
 n 21
 n 22
 n 2 n   
  ... 
C   
a a a in 
i 1
i1
i 1
i2
i 1  c  ,
   i
 n  ... 
 ...  c 
   n
 an1  an 2  ann 
 n n n

  ai1  ai 2  ain 
 i 1 i 1 i 1

 n 
 Example: Synthesizing Procedure - 0
Step 0: Prepare pairwise comparison matrix
Select the Best
Car

Price MPG Comfort Style

Car A Car A Car A Car A

Car B Car B Car B Car B
Car C Car C Car C Car C
Comfor Car A Car B Car C
t
Car A 1 2 8
Car B 1/2 1 6
Car C 1/8 1/6 1
Example: Synthesizing Procedure - 1
Step 1: Sum the values in each column.

Comfort Car A Car B Car C

Car A 1 2 8
Car B 1/2 1 6
Car C 1/8 1/6 1
Column totals 13/8 19/6 15
Example: Synthesizing Procedure - 2
Step 2: Divide each element of the matrix by its column
total.
 All columns in the normalized pairwise comparison
matrix now have a sum of 1.

Comfort Car A Car B Car C

Car A 8/13 12/19 8/15
Car B 4/13 6/19 6/15
Car C 1/13 1/19 1/15
Example: Synthesizing Procedure - 3
Step 3: Average the elements in each row.
 The values in the normalized pairwise comparison matrix have
been converted to decimal form.
 The result is usually represented as the (relative) priority vector.

Comfort Car A Car B Car C Row Avg.

Car A 0.615 0.632 0.533 0.593 0.593

0.341
Car B 0.308 0.316 0.400 0.341  
0.066
Car C 0.077 0.053 0.067 0.066
Total 1.000
Consistency
 An important consideration in terms of the quality of
the ultimate decision relates to the consistency of
judgments that the decision maker demonstrated
during the series of pairwise comparisons.
 It should be realized perfect consistency is very difficult to
achieve and that some lack of consistency is expected to
exist in almost any set of pairwise comparisons.
Consistency
 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.
Consistency Ratio
 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.
Procedure: Estimating Consistency Ratio
Step 1: Multiply each value in the first column of the
pairwise comparison matrix by the relative priority
of the first item considered. Same procedures for
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 max.
   a11   a12   a1n    c1 
        
  a21   a 22   a 2 n    c 2
c1  c2  ...  cn  
 ...  ...  ...   ... 

  an1       
  an 2   ann    cn 
max 
n
Procedure: Estimating Consistency Ratio - 2
Step 4: Compute the consistency index (CI):
λ max  n
CI 
n 1
Where n is the number of items being compared

Step 5: Compute the consistency ratio (CR):

CI
CR 
RI
Where RI is the random index, which is the consistency index of a
randomly generated pairwise comparison matrix. It can be shown
that RI depends on the number of elements being compared and
takes on the following values.
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
Example: Inconsistency
Preferences: If, A  B (2); B  C (6)
Then, A  C (should be 12) (actually 8)

 Inconsistency

Comfort Car A Car B Car C

Car A 1 2 8
Car B 1/2 1 6
Car C 1/8 1/6 1
Example: Consistency Checking - 1
Step 1: Multiply each value in the first column of the pairwise
comparison matrix by the relative priority of the first item
considered. Same procedures for other items. Sum the
values across the rows to obtain a vector of values labeled
“weighted sum.”

1  2 8   0.593 0.682 0.528 1.803

0.593 1 2  0.341  1   0.066 6  0.297   0.341  0.396  1.034 
1 8  1 6 1  0.074 0.057 0.066 0.197
Example: Consistency Checking - 2
Step 2: Divide the elements of the vector of weighted
sums by the corresponding priority value.

1.803 0.593  3.040

1.034 0.341   3.032
   
0.197 0.066 2.985

Step 3: Compute the average of the values computed in

step 2 (max).

3.040  3.032  2.985

λ max   3.019
3
Example: Consistency Checking - 3
Step 4: Compute the consistency index (CI).

λ max  n 3.019  3
CI    0.010
n 1 3 1

Step 5: Compute the consistency ratio (CR).

CI 0.010
CR    0.017  0.10
RI 0.58
 The degree of consistency exhibited in the pairwise
comparison matrix for comfort is acceptable.
Development of Priority Ranking
 The overall priority for each decision
alternative is obtained by summing the product
of the criterion priority (i.e., weight) (with
respect to the overall goal) times the priority
(i.e., preference) of the decision alternative
with respect to that criterion.
 Ranking these priority values, we will have
AHP ranking of the decision alternatives.
Example: Priority Ranking
Step 0A: Other pairwise comparison matrices

Comfort Car A Car B Car C Price Car A Car B Car C

Car A 1 2 8 Car A 1 1/3 ¼
Car B 1/2 1 6 Car B 3 1 ½
Car C 1/8 1/6 1 Car C 4 2 1

MPG Car A Car B Car C Style Car A Car B Car C

Car A 1 1/4 1/6 Car A 1 1/3 4
Car B 4 1 1/3 Car B 3 1 7
Car C 6 3 1 Car C 1/4 1/7 1
Criterion Priority

Criterion Price MPG Comfort Style Priority

Price 1 3 2 2 0.398
MPG 1/3 1 1/4 1/4 0.085
Comfort 1/2 4 1 1/2 0.218

Style 1/2 4 2 1 0.299

Example: Alternatives Ranking
Step 0B: Calculate priority vector for each matrix.

Price MPG Comfort Style Overall ranking

(0.398 (0.085 (0.218) (0.299 of Cars
)  0.087
0.123 )  0.593 ) 
0.265 0.265 
0.320 0.274  0.341 0.655 0.421 
Car A          
Car B 0.557 0.639  0.066 0.080 0.314 

Car C

Overall car A priority  0.398 (0.123)  0.085 (0.087)  0.218 (0.593)

 0.299 (0.265)
 0.265
TOPSIS

Technique of Order Preference

by Similarity to Ideal Solution

33
 TOPSIS METHOD
 Technique of Order Preference by Similarity
to Ideal Solution
 This method considers two types of attributes
or criteria

 Benefit attributes/criteria
 Cost attributes or criteria

34
 TOPSIS METHOD
 In this method two artificial alternatives are
hypothesized:

 Ideal alternative: the one which has the best level

for all attributes considered.
 Negative ideal alternative: the one which has the
worst attribute values.

 TOPSIS selects the alternative that is the closest to

the positive ideal solution and farthest from
negative ideal alternative.
35
 Input to TOPSIS
 TOPSIS assumes that we have m alternatives
(options) and n attributes/criteria and we have the
score of each option with respect to each criterion.

 Let xij score of option i with respect to criterion j

We have a matrix X = (xij) mn matrix.
 Let J be the set of benefit attributes or criteria
(more is better)
 Let J' be the set of negative attributes or criteria
(less is better)
36
 Steps of TOPSIS
 Step 1: Construct normalized decision
matrix.
 This step transforms various attribute
dimensions into non-dimensional attributes,
which allows comparisons across criteria.

 Normalize scores or data as follows:

xij
rij  ; for i  1,..., m; j  1,..., n
 ij
i
x 2

37
 Steps of TOPSIS
 Step 2: Construct the weighted normalized
decision matrix.
 Assume we have a set of weights for each
criteria wj for j = 1,…n.
 Multiply each column of the normalized
decision matrix by its associated weight.
 An element of the new matrix is:

vij = wj rij
38
 Steps of TOPSIS
 Step 3: Determine the ideal and negative ideal
solutions.

 Ideal solution.
A* = { v1* , …, vn*}, where
vj* ={ max (vij) if j  J ; min (vij) if j  J' }
i i

 Negative ideal solution.

A' = { v1' , …, vn' }, where
v' = { min (vij) if j  J ; max (vij) if j  J' }
i i

39
 Steps of TOPSIS

 Step 4: Calculate the separation measures for each

alternative.

 The separation from the ideal alternative is:

Si * = [  (vj*– vij)2 ] ½ i = 1, …, m
j

 Similarly, the separation from the negative ideal

alternative is:
S'i = [  (vj' – vij)2 ] ½ i = 1, …, m
j

40
 Steps of TOPSIS
 Step 5: Calculate the relative closeness to the
ideal solution Ci*

Ci* = S'i / (Si* +S'i ) , 0  C i*  1

Select the option with Ci* closest to 1.

41
 Applying TOPSIS Method to
Example

Weights 0.299 0.218 0.085 0.398

Style Comfort MPG Price
(10 point (10 point (10 point in Rs.
Criteria scale) scale) scale) (Lakh)
Civic 7 9 9 8
Saturn 8 7 8 7
Ford 9 6 8 9
Mazda 6 7 8 6

42
 Applying TOPSIS to Example
 m = 4 alternatives (car models)
 n = 4 attributes/criteria

 xij = score of option i with respect to criterion j

X = {xij} 44 score matrix.
 J = set of benefit attributes: style, reliability, fuel
economy (more is better)
 J' = set of negative attributes: cost (less is better)

43
 Steps of TOPSIS
 Step 1(a): calculate (x2ij )1/2 for each column

Weights 0.299 0.218 0.085 0.398

Style Comfort MPG Price
(10 point (10 point (10 point in Rs.
Criteria scale) scale) scale) (Lakh)
Civic 7 9 9 8
Saturn 8 7 8 7
Ford 9 6 8 9
Mazda 6 7 8 6
(x2ij )1/2 15.165 14.662 16.522 15.165
 Steps of TOPSIS
 Step 1 (b): divide each column by (x2ij )1/2 to get rij

Weights 0.299 0.218 0.085 0.398

Style Comfort MPG Price
Civic 0.461566 0.613794834 0.544704834 0.527504
Saturn 0.527504 0.477395982 0.484182074 0.461566
Ford 0.593442 0.409196556 0.484182074 0.593442
Mazda 0.395628 0.477395982 0.484182074 0.395628
 Steps of TOPSIS
 Step 2 (b): multiply each column by wj to get vij.

Style Comfort MPG Price

Civic 0.138008 0.133807 0.0463 0.209947
Saturn 0.157724 0.104072 0.041155 0.183703
Ford 0.177439 0.089205 0.041155 0.23619
Mazda 0.118293 0.104072 0.041155 0.15746

46
 Steps of TOPSIS
 Step 3 (a): determine ideal solution A*.
A* = {0.177439, 0.133807, 0.0463, 0.15746}

Style Comfort MPG Price

Civic 0.138008 0.133807 0.0463 0.209947
Saturn 0.157724 0.104072 0.041155 0.183703
Ford 0.177439 0.089205 0.041155 0.23619
Mazda 0.118293 0.104072 0.041155 0.15746

47
 Steps of TOPSIS
 Step 3 (a): find negative ideal solution A'.
A' = {0.118293, 0.089205, 0.041155, 0.23619}

Style Comfort MPG Price

Civic 0.138008 0.133807 0.0463 0.209947
Saturn 0.157724 0.104072 0.041155 0.183703
Ford 0.177439 0.089205 0.041155 0.23619
Mazda 0.118293 0.104072 0.041155 0.15746

48
 Steps of TOPSIS
 Step 4 (a): determine separation from ideal solution A* = {0.059,
0.244, 0.162, 0.080} Si* = [  (vj*– vij)2 ] ½ for each row j

Style Comfort MPG Price Si*

Civic 0.001555 0 0 0.002755 0.065648

Saturn 0.000389 0.000884151 2.64661E-05 0.000689 0.044587

Ford 0 0.001989352 2.64661E-05 0.006198 0.090632

Mazda 0.003498 0.000884151 2.64661E-05 0 0.066399

49
 Steps of TOPSIS
 Step 4 (b): find separation from negative ideal solution A' = {0.040, 0.164,
0.144, 0.118}
Si' = [  (vj'– vij)2 ] ½ for each row j

Style Comfort MPG Price Si'

Civic 0.000389 0.001989363 2.64701E-05 0.000689 0.055617

Saturn 0.001555 0.000221037 0 0.002755 0.06731

Ford 0.003498 0 0 0 0.059146

Mazda 0 0.000221037 0 0.006198 0.080121

50
 Steps of TOPSIS
 Step 5: Calculate the relative closeness to
the ideal solution Ci* = S'i / (Si* +S'i )

S'i /(Si*+S'i) Ci*

C Rank
Civic 0.458640239 3
Saturn 0.601534911 1
Ford 0.394891034 4
Mazda 0.546826075 2
51