You are on page 1of 16

Analytic Hierarchy Process (AHP)

 Multi Criteria decision making method developed by Prof.


Thomas L. Saaty
 Provides a comprehensive and rational framework for
structuring and analysing complex decision problems
 Most useful for complex problems, especially those with
high stakes, involving human perceptions and judgments,
whose decisions have long-term consequences
Analytic Hierarchy Process (AHP)
 It decomposes a decision problem into a hierarchy of more
easily comprehended sub-problems, which are analysed
independently
 Capable of using
◦ multiple, diverse criteria
◦ qualitative as well as quantitative information
 Best method when important elements of the decision are
difficult to quantify or compare
 Ability to assign numerical values for priorities
How AHP works?
1. The problem is modelled as a hierarchy containing the
decision goal, the alternatives for reaching it, and the
criteria for evaluating the alternatives.
2. Priorities are established among the elements by making a
series of judgments based on ‘pairwise comparisons’ of the
elements.
3. These judgments are synthesized to yield a set of overall
priorities for the hierarchy.
4. The consistency of judgments is checked through iterations.
5. A final decision is taken based on the results.
How AHP works?
Example
 Suppose you want to buy a new car
 Objective- Selecting a car
 Criteria- Style, Reliability, and
Fuel economy
 Alternatives:
- Hyundai i20
- Honda Amaze
- Chevrolet Sail sedan
- Maruti Suzuki Swift
Example
S e le c t in g
a N ew C ar

S t y le R e lia b ilit y FuelE conom y


- i20 - i20 - i20
- Amaze - Amaze - Amaze
- Sail - Sail - Sail
- Swift - Swift - Swift
Determining weights
 Let’s say you think reliability is the most important criterion
followed by style and fuel economy

 When asked to elaborate using “pair-wise comparisons”, you


say:

[1:Equal, 3:Moderate, 5:Strong, 7:Very strong, 9:Extreme]

◦ Reliability is 2 times as important as style

◦ Reliability is 4 times as important as fuel economy

◦ Style is 3 times as important as fuel economy


Eigen Matrix
Style Reliability Fuel Economy

Style 1/1 1/2 3/1

Reliability 2/1 1/1 4/1

Fuel Economy 1/3 1/4 1/1

 Using pair-wise comparisons, the relative importance of


one criterion over another can be expressed
 For n criterion, we will have a n x n matrix
Eigen vectors
 To solve for Eigen vectors, Iterate
◦ Take successive squared powers of matrix
◦ Normalize the row sums
 Until difference between successive
iterations satisfies a pre-specified
condition (like same till three places of
decimal etc.)
 Several other methods are available to
find these Eigen vectors
Eigen vectors
1 0.5 3 squared
3.0 1.75 8.0
2 1 4 5.3332 3.0 14.0
0.333 0.25 1.0 1.1666 0.6667 3.0
Normalized
Row sums Row sums
12.75 0.3194
22.3332 0.5595
4.8333 0.1211
39.9165 1.0
0.3196
• New iteration gives normalized row sums as 0.5584
0.1220

• Since we needed values same till three decimal places, we can go for
another iteration using the same process
Assigning weights
S e le c t in g
a N ew C ar
1 .0

S t y le R e lia b ilit y FuelE conom y


.3 1 9 6 .5 5 8 4 .1 2 2 0

 Now that we have ranked criteria, we move on to


ranking alternatives
 Again, we take your judgement to form a matrix
Ranking Alternatives
Style i20 Amaze Sail Swift Eigenvector
i20 1/1 1/4 4/1 1/6 .1160
Amaze 4/1 1/1 4/1 1/4 .2470
Sail 1/4 1/4 1/1 1/5 .0600
Swift 6/1 4/1 5/1 1/1 .5770

Reliability
i20 Amaze Sail Swift
i20 1/1 2/1 5/1 1/1 .3790
.2900
Amaze 1/2 1/1 3/1 2/1
.0740
Sail 1/5 1/3 1/1 1/4
.2570
Swift 1/1 1/2 4/1 1/1
Ranking Alternatives
Fuel Economy Kms/litre Normalized

i20 15 .3010
Amaze 12 .2390
Sail 11 .2120
Swift 13 .2480

51 1.0
 For fuel economy, we have quantitative information
and we normalize it directly as shown
Final Hierarchical Tree
S e le c t in g
a N ew C ar
1 .0

S t y le R e lia b ilit y FuelE conom y


.3 1 9 6 .5 5 8 4 .1 2 2 0

- i20 .1160 - i20 .3790 - i20 .3010


- Amaze .2470 - Amaze .2900 - Amaze .2390
- Sail .0600 - Sail .0740 - Sail .2120
- Swift .5770 - Swift .2570 - Swift .2480
The Decision?
Style Reliability Fuel economy
i20 .1160 .3790 .3010 .3196 .3060
Amaze .2470 .2900 .2390 * = .2720
.5584
Sail .0600 .0740 .2120 .0940
Swift .1220 .3280
.5770 .2570 .2480

The analysis shows that Swift is the decision


maker’s preferred choice
Some real-life applications
 Quantifying the overall quality of software systems (By
Microsoft Corporation)
 Selecting university faculty (University of Pennsylvania)
 Deciding where to locate offshore manufacturing plants
(University of Cambridge)
 Assessing risk in operating cross-country petroleum
pipelines (By American Society of Civil Engineers)
 Deciding how best to manage U.S. watersheds (By U.S.
Department of Agriculture)

You might also like