Sunteți pe pagina 1din 34

DECISION MODELING WITH

MICROSOFT EXCEL
Chapter 12
Multi-Objective Decision Making

Copyright 2001
Prentice Hall

ANALYTICAL HIERARCHY
PROCESS
This section deals with the real-world topic of making a
decision when there are multiple objectives or criteria to
consider. For example:
Choosing which employment offer to accept.
Picking which computer (or car, etc.) to buy.
Deciding which new product to launch first.
Selecting a site for a new restaurant, hotel, etc.
Rating the best cities in which to live.
Choosing a new software package for your
company.

A simple way to attack such a decision would be to


assign weights to each of the criteria that were to be
considered in making the decision.
Then, rank each decision alternative on a scale from 1
(worst) to 10 (best).
Finally, you would multiply the weights times the
rankings for each criterion and sum them up.
The alternative with the highest score would be the most
preferred.

For example, you are in charge of purchasing the next


computer for the office. You have to choose between the
following three computers:
1. Model A runs an AMD K6-II chip at 400 MHz
2. Model B runs a Celeron chip at 333 MHz
3. Model C runs a Pentium II chip at 450 MHz
The important criteria and their weights are:
Criteria
Price
Speed
Hard-disk Size
Warranty/Support

Weight
50%
15%
20%
15%

Now, rank each of the three models on these four criteria.


Rank them on a scale from 1 to 10 as described earlier.

=SUM(C4:C7)

=SUMPRODUCT($C$4:$C$7,E4:E7)

Model B has the highest weighted score and thus would


be the best computer to purchase.
5

This approach is quite simplistic and there are


difficulties in setting the ranking scales on such different
criteria.
Analytic hierarchy process (AHP) also uses a weighted
average approach idea, but it uses a method for
assigning ratings (or rankings) and weights that is
considered more reliable and consistent.
(AHP) is based on pairwise comparisons between the
decision alternatives on each of the criteria.
Then, a similar set of comparisons are made to
determine the relative importance of each criterion and
thus produces the weights.
6

Analytic Hierarchy Process


Multiple criteria
quantitative
qualitative, intangible, subjective
provides measures of judgement consistency
derives priorities among criteria and alternatives
user-friendly pair-wise comparisons

Using AHP
1. Decompose the problem into a hierarchy
2. Make pairwise comparisons and
establish priorities among the elements in
the hierarchy
3. Synthesise the results (to obtain the
overall ranking of alternatives w.r.t. goal)
4. Evaluate the consistency of judgement

The basic procedure is as follows:


1. Develop the ratings for each decision alternative
for each criterion by
developing a pairwise comparison matrix for
each criterion
normalizing the resulting matrix
averaging the values in each row to get the
corresponding rating
calculating and checking the consistency
ratio

2. Develop the weights for the criteria by


developing a pairwise comparison matrix for
each criterion
normalizing the resulting matrix
averaging the values in each row to get the
corresponding rating
calculating and checking the consistency
ratio
3. Calculate the weighted average rating for each
decision alternative. Choose the one with the
highest score.
10

Consider the following example:


Sleepwell Hotels is looking for some help in selecting the
best revenue management software package from
among several vendors. The director of revenue
management for this chain of hotels has been given this
task.
Three vendors have been identified whose software
meets the following basic needs:
Revenue Technology Corporation (RTC)
PRAISE Strategic Solutions (PSS)
El Cheapo (EC)
11

The important criteria are:


1. The total cost of the installed system
2. The follow-up service provided over the coming
year
3. The sophistication of the underlying math
engines
4. The amount of customization for Sleepwell

12

The first step in the AHP procedure is to make pairwise


comparisons between the vendors for each criterion.
Here is the standard scale for making these
comparisons:
RATING
1
3
5
7
9

DESCRIPTION
Equally preferred
Moderately preferred
Strongly preferred
Very strongly preferred
Extremely strongly preferred

Values 2, 4, 6, or 8 may also be assigned and represent


preferences halfway between the integers on either side.
13

Start with the total cost criterion and generate the


following data in a spreadsheet:

=1/D4

=1/C4
=1/D5

The vendor in the row is being compared to the vendor in


the column.
A value between 1 and 9 indicates that the vendor in the
row is preferred to the vendor in the column.
If the vendor in the column is preferred to the vendor14 in
the row, then the inverse of the rating is given.

The next step is to normalize the matrix. This is done by


totaling the numbers in each column.
Each entry in the column is then divided by the column
sum to yield its normalized score.

=SUM(B4:B6)
=B4/B$8

The average is
calculated for the
Total Cost
criterion.
=AVERAGE(B12:D12)
Highest
average
score
15

Now, calculate the consistency ratio and check its value.


The purpose for doing this is to make sure that the
original preference ratings were consistent.
There are 3 steps to arrive at the consistency ratio:
1. Calculate the consistency measure for each
vendor.
2. Calculate the consistency index (CI).
3. Calculate the consistency ratio (CI/RI where RI
is a random index).
To calculate the consistency measure, we can take
advantage of Excels matrix multiplication function
=MMULT().
16

Multiply the average rating for each vendor times the


scores in the first row one-at-a-time, sum these products
up and divide this sum by the average rating for the first
vendor.

=MMULT(B4:D4,$E$12:$E$14)/E12
=(AVERAGE(F12:F14)-3)/2
Provided by
AHP (see next
slide)
=F16/F18)

17

Approximation of the Consistency Index


1. Multiply each column of the pairwise
comparison matrix by the corresponding
weight.
2. Divide of sum of the row entries by the
corresponding weight.
3. Compute the average of the values from step 2,
denote it by Lmax.
L max n
4. The approximate CI is

n 1
18

N
2
3
4
5
6
7
8
9
10

RANDOM INDEX
0.00
0.58
0.90
1.12
1.24
1.32
1.41
1.45
1.51

Random Index (RI)


the CI of a randomly-generated pairwise comparison matrix
19

If we are perfectly consistent, then the consistency


measures will equal n and therefore, the CIs will be equal
to zero and so will the consistency ratio.
If this ratio is very large (Saaty suggests > 0.10), then we
are not consistent enough and the best thing to do is go
back and revise the comparisons.
Now, continue for the other three criteria. You can easily
do this by copying the Total Cost sheet into three other
sheets (Service, Sophistication, and Custom) and
then simply changing the pairwise comparisons.

20

Consistency ratio for Service.

21

Consistency ratio for Sophistication.

22

Consistency ratio for Customization.

23

In all three cases, the CR value ranges from 0.0 to 0.047


which means that we are being consistent.
Note also that PSS is the winner on the Service criterion,
RTC and PSS are tied for the best in terms of
Sophistication, and PSS is considered the best on
Customization.
All of this work concludes the first step in the procedure.
The next step is to use similar pairwise comparisons to
determine the appropriate weights for each of the
criteria.
The process is the same in that we make comparisons,
except that now we make the comparisons between the
criteria not the vendors.
24

Consistency ratio for weights on criterion.


=1/C4
=1/D4
=1/E4

=1/D5
=1/E5

=SUM(B4:B7)

=1/E6

=AVERAGE(B12:E12)
=MMULT(B4:E4,$F$12:$F$15)/F12

=B4/B$8
=(AVERAGE(G12:G15)-4)/3
=G16/G18)
25

The final step is to calculate the weighted average


ratings of each decision alternative and use the results
to decide from which vendor to purchase the software.
=WEIGHTS!F12
=TOTAL
=TOTAL
COST!E12
COST!E13
=TOTAL COST!E14
=SERVICE!E12
=SERVICE!E13
=SERVICE!E14
=SOPHISTICATION!E12
=SOPHISTICATION!E13
=SOPHISTICATION!E14
=CUSTOM!E12
=CUSTOM!E13
=CUSTOM!E14

=SUMPRODUCT($B$3:$B$6,C3:C6)
These results are pulled from all the other worksheets.
From these results, we find that RTC barely edges out
26
PSS for the software contract.

The mathematics of AHP


Suppose we already know the weights [w1, w2, w3, . . . wn] of the n
criteria and we form the following n x n pairwise-ratio matrix:

27

This pairwise-ratio matrix A and the vector of weights satisfy the


following equation:

28

This equation is of the form:


A w = w
So w is an eigenvector of matrix A corresponding
to eigenvalue
(In fact, is the only non-zero eigenvalue, and w the
unique eigenvector.)

Now, if we only know A, but not w, we can find


what w is by solving for the eigenvalues and
eigenvectors of A.
29

Back to AHP ...


If :

we use a continuous scale instead of a 9-point


scale, and,
more importantly, our judgement is consistent,
then the pairwise comparison matrix is exactly of
the form A
and the weights for the criteria are given by the
eigenvector corresponding to eigenvalue .
30

Computing the weights for AHP


Eigenvector Method:
1. Find largest eigenvalue of the pairwise comparison
matrix
2. Find corresponding eigenvector
Approximate Method:
1. Normalise each column (i.e. divide each entry by its
column total)
2. The average values of row i in the normalised matrix
is the estimate for weight i.
31

Consistency Index
reflects the consistency of
ones judgement

CI = . max n .
n 1

Random Index (RI)


the CI of a randomlygenerated pairwise
comparison matrix

Tabulated by size of matrix:


. n
2
3
4
5
6
7
8
9
10

RI .
0.0
0.58
0.90
1.12
1.24
1.32
1.41
1.45
1.51

32

Consistency Ratio
CR = CI / RI
In practice, a CR of 0.1 or below is considered
acceptable.
Any higher value at any level indicate that the
judgements warrant re-examination.
33

AHP - Summary
Issues:

Interdependence
Interval Judgements
Is a multiplicative scale appropriate?

Easy to use
Intuitive?
General framework
Widely used

Cost/Benefit Analysis
Vendor Selection
Strategic Planning
34

S-ar putea să vă placă și