1

Multi-Criteria Decision Making

MCDM Approaches

Introduction
Zeleny (1982) opens his book Multiple Criteria Decision Making with a statement:
It has become more and more difficult to see the world around us in a unidimensional way and to use only a single criterion when judging what we see

Introduction

Many public sector problems and even private decision involve multiple objectives and goals. As an example: Locating a nuclear power plant involves objectives such as: Safety Health Environment Cost
3

Examples of Multi-Criteria Problems

In a case study on the management of R&D research (Moore et. al 1976), the following objectives have been identified:
Profitability Growth and diversity of the product line Increased market share Maintained technical capability Firm reputation and image Research that anticipates competition

Examples of Multi-Criteria Problems

In determining an electric route for power transmission in a city, several objectives could be considered:
Cost Health Reliability Importance of areas
5

Examples of Multi-Criteria Problems

In selecting a major at KFUPM, several objectives can be considered. These objectives or criteria include:
Job market upon graduation Job pay and opportunity to progress Interest in the major Likelihood of success in the major Future job image Parent wish
6

Examples of Multi-Criteria Problems

Wife selection problem. This problem is a good example of multi-criteria decision problem. Criteria include:
Religion Beauty Wealth Family status Family relationship Education
7

Approaches For MCDM

Several approaches for MCDM exist. We will cover the following:

Weighted score method ( Section 5.1 in text book). TOPSIS method Analytic Hierarchy Process (AHP) Goal programming ?
8

Weighted score method

Determine the criteria for the problem Determine the weight for each criteria. The weight can be obtained via survey, AHP, etc. Obtain the score of option i using each criteria j for all i and j Compute the sum of the weighted score for each option .
9

Weighted score method

In order for the sum to make sense all criteria scale must be consistent, i.e., More is better or less is better for all criteria

Example: In the wife selection problem, all criteria (Religion, Beauty, Wealth, Family status, Family relationship, Education) more is better If we consider other criteria (age, dowry) less is better
10

Weighted score method

Let Sij score of option i using criterion j wj weight for criterion j Si score of option i is given as:
Si = wj Sij
j

The option with the best score is selected.

11

Weighted Score Method

The method can be modified by using U(Sij) and then calculating the weighted utility score. To use utility the condition of separability must hold. Explain the meaning of separability: U(Si) = wj U(Sij) U(Si) U( wj Sij)
12

Example Using Weighted Scoring Method

Objective
Selecting a car

Criteria
Style, Reliability, Fuel-economy

Alternatives
Civic Coupe, Saturn Coupe, Ford Escort, Mazda Miata
13

Weights and Scores

Weight 0.3 0.4 0.3 Si

Style
Civic Saturn Ford 7 8 9

Reliability
9 7 6

Fuel Eco.
9 8 8

8.4 7.6 7.5 7.0

14

Mazda

TOPSIS METHOD

Technique of Order Preference by Similarity to Ideal Solution This method considers three types of attributes or criteria
Qualitative benefit attributes/criteria Quantitative benefit attributes Cost attributes or criteria
15

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 ideal solution and farthest from negative ideal alternative.
16

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) mn 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)
17

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: rij = xij/ (x2ij) for i = 1, , m; j = 1, , n
i
18

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
19

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
20

Steps of TOPSIS

Step 4: Calculate the separation measures for each alternative. The separation from the ideal alternative is: Si * = [ (vj* vij)2 ] i = 1, , m
j

Similarly, the separation from the negative ideal alternative is: S'i = [ (vj' vij)2 ] i = 1, , m
j
21

Steps of TOPSIS

Step 5: Calculate the relative closeness to the ideal solution Ci* Ci* = S'i / (Si* +S'i ) , 0 Ci* 1

Select the option with Ci* closest to 1. WHY ?

22

Applying TOPSIS Method to Example

Weight 0.1 Style Civic Saturn Ford Mazda 7 0.4 Reliability 9 0.3 0.2 Fuel Eco. Cost 9 8

8
9 6

7
6 7

8
8 8

7
9 6

23

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} 44 score matrix. J = set of benefit attributes: style, reliability, fuel economy (more is better) J' = set of negative attributes: cost (less is better)
24

Steps of TOPSIS

Step 1(a): calculate (x2ij )1/2 for each column

Style Rel. Fuel
Civic Saturn Ford 49 64 81 36 230 81 49 36 49 215 81 64 64 64 273 Cost 64 49 81 36 230
25

Mazda

xij2
i

(x2)1/2

15.17 14.66 16.52 15.17

Steps of TOPSIS

Step 1 (b): divide each column by (x2ij )1/2 to get rij

Style Civic Saturn Ford Mazda Rel. Fuel Cost

0.46
0.53 0.59 0.40

0.61
0.48 0.41 0.48

0.54
0.48 0.48 0.48

0.53
0.46 0.59 0.40
26

Steps of TOPSIS

Step 2 (b): multiply each column by wj to get vij.

Style Civic Saturn Ford Mazda Rel. Fuel Cost

0.046 0.244 0.162 0.106

0.053 0.192 0.144 0.092 0.059 0.164 0.144 0.118 0.040 0.192 0.144 0.080
27

Steps of TOPSIS

Step 3 (a): determine ideal solution A*. A* = {0.059, 0.244, 0.162, 0.080}
Style Civic Saturn Rel. Fuel Cost

0.046 0.244 0.162 0.106 0.053 0.192 0.144 0.092

Ford
Mazda

0.059 0.164 0.144 0.118

0.040 0.192 0.144 0.080
28

Steps of TOPSIS

Step 3 (a): find negative ideal solution A'. A' = {0.040, 0.164, 0.144, 0.118}
Style Civic Saturn Ford Mazda Rel. Fuel

Cost

0.046 0.244 0.162 0.106 0.053 0.192 0.144 0.092

0.059 0.164 0.144 0.118

0.040 0.192 0.144 0.080
29

Steps of TOPSIS

Step 4 (a): determine separation from ideal solution A* = {0.059, 0.244, 0.162, 0.080} Si* = [ (vj* vij)2 ] for each row
j

Style Civic (.046-.059)2

Rel. (.244-.244)2

Fuel (0)2

Cost
(.026)2

Saturn (.053-.059)2 Ford (.053-.059)2

(.192-.244)2 (-.018)2 (.012)2 (.164-.244)2 (-.018)2 (.038)2 (.192-.244)2 (-.018)2 (.0)2

30

Mazda (.053-.059)2

Steps of TOPSIS

Step 4 (a): determine separation from ideal solution Si*

(vj*vij)2 Civic Saturn 0.000845 0.003208 Si* = [ (vj* vij)2 ] 0.029 0.057

Ford
Mazda

0.008186
0.003389

0.090
0.058
31

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 ] for each row
j

Style Civic (.046-.040)2

Rel. (.244-.164)2

Fuel (.018)2

Cost
(-.012)2 (-.026)2 (0)2 (-.038)2
32

Saturn (.053-.040)2 Ford (.053-.040)2

(.192-.164)2 (0)2 (.164-.164)2 (0)2 (.192-.164)2 (0)2

Mazda (.053-.040)2

Steps of TOPSIS

Step 4 (b): determine separation from negative ideal solution Si'

(vj'vij)2 Civic Saturn 0.006904 0.001629 Si' = [ (vj' vij)2 ] 0.083 0.040

Ford
Mazda

0.000361
0.002228

0.019
0.047
33

Steps of TOPSIS

Step 5: Calculate the relative closeness to the ideal solution Ci* = S'i / (Si* +S'i )
S'i /(Si*+S'i) Civic Saturn 0.083/0.112 0.040/0.097 Ci* 0.74 0.41 BEST

Ford
Mazda

0.019/0.109
0.047/0.105

0.17
0.45
34