Sunteți pe pagina 1din 7

# 14th Session: Multiple criteria Decisions

Learning Objectives

Goal Programming Methods
Analytical Hierarchy Process

1. ntroduction
2. The Conception of Goal Programming
3. Examples of Goal Programming
4. Analytic Hierarchy Process (AHP)
5. Examples of AHP

Learning Contents
ntroduction
So far, we only consider a single criterion in decision making
ex) maximize profit, minimize cost, minimize time
However, more than one criterion may be simultaneously considered.
ex) When searching a location for a new manufacturing land
Decision criteria -> The cost for land and construction of plant
Availability of transportation from plant to firm
Hiring and retaining employees
Approaches for Solving multi criteria decision making
Goal Programming (GP)
Handles multiple criteria situations within the general framework of Linear
Programming
Analytic Hierarchy Process (AHP).
Allows the users to arrive at a prioritized ranking of the decision
alternatives.
Nicolo nvestment - llustrative Example for Goal Programming
A client has \$80,000 to invest and, as an initial strategy, would like the investment
portfolio restricted to two stocks:
Stock Price/share Estimated annual
return/share
Risk index/share
U.S. Oil \$25 \$3
(rate of return: 12%)
0.50
Hub Properties \$50 \$5
(rate of return: 10%)
0.25
The risk index per share, 0.5 for US Oil and 0.25 for Hub Properties, is a rating Nicolo
assigned to measure the relative risk of the two measurements. Higher risk index values
imply greater risk. Hence Nicolo judged US Oil to the risker investment.
Suppose Nicolo chooses a portfolio that invests all \$80,000 in US Oil, Nicolo could
purchase \$80,000/\$25 = 3,200 shares of US oil and the portfolio would have a risk index
of 3200(0.5) = 1600. f no share of either stock, the portfolio will have no risk, but also no
return.
Nicolo's client would like to find a portfolio with a maximum total risk index of 700 to
obtain an annual return of at least \$9,000. He has \$80,000 for investment. His top-
priority goal is to restrict the risk: that is keeping the portfolio risk index at 700 or less.
Based on this statement, the goals are the follows
Goal1 (priority level 1) : Find a portfolio that has a risk index of 700 or less
Goal2 (priority level 2): Find a portfolio that will provide an annual return of at
least \$9000.
Developing Problem of Nicolo nvestment
Decision Variables
U = number of shares of US Oil purchased
H = number of shares of Hub Properties purchased
Constraints
Handled in the same way as in linear programming
Fund available
0 + B 8,
Goal Equation for each goal
Each goal is represented as function with deviation variables.
Deviation Variable
Allow for the possibility of not meeting the target value exactly.
Goal equations and deviation variables (Goal1)
Portfolio risk index, (.0 +.B) may be less than, equal to, or greater than the
target value 700, which can be represented as follows

.0 + .B = +u
1
+
- u
1
-

u
1
+
(Deviation Variable): the amount by which the portfolio risk index exceeds the
target value of 700
u
1
-
(Deviation Variable): the amount by which the portfolio risk index is less than the
target value of 700

Ex: f U = 2000 and H =0, the portfolio index is 1000.-> u
1
+
= anu u
1
-
=
This portfolio index is greater than target value, 700.

Ex: f U = 0 and H =1000, the portfolio index is 250.-> u
1
+
= anu u
1
-
=
This solution, 450 is satisfied with goal1, since it is less than target value 700.

Rewrite the goal equation
Goal equation for goal 1
.0 + .B - u
1
+
+ u
1
-
=
Two parts of goal equation
A function that defines the amount of goal achievement in terms of decision
variables (e.g. .0 + .B)
Deviation variables representing the difference between the target value for the
goal and the level achieved.

Goal equations and deviation variables (Goal2)
Annual return of investment (0 + B) may be also less than, equal to, or greater
than the target value 9000, which can be represented as follows

0 +B = 9 +u

+
-u

-

u
2
+
: the amount by which the annual return for the portfolio exceeds the target
value of 9000
u
2
-
: the amount by which the annual return for the portfolio is less than the target
value of 9000
or
0 +B -u

+
+u

-
= 9
Solution process for Goal Programming
Objective function of goal programming
Minimizing a function of the deviation variables
Process
1) The priority level 1 linear programming is solved.
Min u
1
+

s. t.
0 +B 8, : Fund Available
.0 + .B -u
1
+
+u
1
-
= : First Goal (Portfolio risk index)
0 + B - u

+
+u

-
= 9 : Second Goal (Annual return of investment)
0, B, u

+
, u

-
, u

+
, u

-

2) The priority level 2 linear programming is solved with maintaining the
achievement of first goal

Min u
2
-

s. t.
0 +B 8, : Fund Available
.0 + .B -u
1
+
+u
1
-
= : First Goal (Portfolio risk index)
0 + B - u

+
+u

-
= 9 : Second Goal (Annual return of investment)
u
1
+
= : Maintain achievement of first goal
0, B, u

+
, u

-
, u

+
, u

-

Goal Programming in Excel (Sketch)

Summary of the procedure to develop a goal programming model
Step1) dentify the goals and any constraints that reflect resource capabilities or
other restrictions that may prevent achievement of the goal
Step2) Determine the priority level of each goal
Step3) Define the decision variables
Step4) Formulate the constraints in the usual linear programming fashion
Step5) For each goal, develop a goal equation. Deviation variables u
1
+
, u
1
-
are
included in each goal equation.
Step6) Develop the objective function in terms of minimizing a prioritized function of
the decision variables.

Car Selection Model illustrative example for Analytic Hierarchy Process (AHP)
We consider a car purchasing decision problem. After a preliminary analysis of the
makers and models of several used cars, the list of decision alternatives included
the three cars: Honda Accord, Saturn, and a Chevrolet Cavalier.
The decision criteria were as follows
Price
Miles per gallon(MPG)
Comfort
Style
Developing hierarchy for the car selection problem

Establishing priorities using AHP
How AHP determine priorities for each of the following:
How the four criteria contribute to the overall goal of selecting the best car.
How the three cars compare using the Price criterion
How the three cars compare using the MPG criterion
How the three cars compare using the Comfort criterion
How the three cars compare using the Style criterion
Procedure for AHP
Pairwise Comparison,
Make pairwise comparison matrix
Synthesization
Other pairwise comparisons
Develop an overall priority ranking
Pairwise Comparison
Statement that how important each criteria is relative to each other criterion when
the criteria are compared two at a time (pariwise).
Price compared to MPG
Price compared to Comfort
Price compared to Style
MPG compared to Comfort
MPG compared to Style
Comfort compared to Style
Comparison Scale
Verbal Judgment Numerical Rating
Extremely more important
9
8
Very strongly more important
7
6
Strongly more important
5
4
Moderately more important
3
2
Equally important 1

Example of Pairwise Comparison
Pairwise Comparison More important Criterion How much mportant Rating
Select the best car
Price
Accord
Saturn
Cavalier
MPG
Accord
Saturn
Cavalier
Comfort
Accord
Saturn
Cavalier
Style
Accord
Saturn
Cavalier
Price-MPG Price Moderately 3
Price-Comfort Price Equally to moderately 2
Price-Style Price Equally to moderately 2
MPG-Comfort Comfort Moderately to strongly 4
MPG-Style Style Moderately to strongly 4
Comfort-Style Style Equally to moderately 2

Pairwise Comparison Matrix
Each of the numeric rating must be entered into the pariwise comparison matrix.
Price MPG Comfort Style
Price 3 2 2
MPG
Comfort 4
Style 4 2
Diagonal elements are all equal to 1 because the elements are comparing each
criterion to itself.
Price MPG Comfort Style
Price 1 3 2 2
MPG 1
Comfort 4 1
Style 4 2 1
All the Remains is to complete the entries for the remaining cells of the matrix. Each
reverse comparison has reciprocal value.
Ex: Price-MPG is 3, then MPG-Price is 1/3. Comfort-MPG is 4, then MPG-Comfort
is 1/4.
Price MPG Comfort Style
Price 1 3 2 2
MPG 1/3 1 1/4 1/4
Comfort 1/2 4 1 12
Style 1/2 4 2 1
Synthesization
Sum the values in each column of the pairwise comparison matrix.
Price MPG Comfort Style
Price 1 3 2 2
MPG 1/3 1 1/4 1/4
Comfort 1/2 4 1 12
Style 1/2 4 2 1
Sum 2.333 12.000 5.250 3.750

Divide each element in the pairwise comparison matrix by its column total: the
resulting matrix is referred to the normalized pairwised comparison matrix
Price MPG Comfort Style
Price 0.429 0.250 0.381 0.533
MPG 0.143 0.083 0.048 0.067
Comfort 0.214 0.333 0.190 0.133
Style 0.214 0.333 0.381 0.267

Compute the average of the elements in each row of the normalized pairwise
comparison matrix.
Price MPG Comfort Style priority
Price 0.429 0.250 0.381 0.533 0.398
MPG 0.143 0.083 0.048 0.067 0.085
Comfort 0.214 0.333 0.190 0.133 0.218
Style 0.214 0.333 0.381 0.267 0.299
Other Pairwise Comparisons for the car selection problem
The following pairwise comparisons are needed for each criterion.
The Accord compared to the Saturn
The Accord compared to the Cavilier
The Saturn compared to the Caviller
Ex: For price, all pairwise comparison are needed.
The Results of pairwise comparison for the cars
Price
Accord Saturn Cavalier
Accord 1 1/3 1/4
Saturn 3 1 1/2
Cavalier 4 2 1
MPG
Accord Saturn Cavalier
Accord 1 1/4 1/6
Saturn 4 1 1/3
Cavalier 6 3 1
Comfort
Accord Saturn Cavalier
Accord 1 2 8
Saturn 1/2 1 6
Cavalier 1/8 1/6 1
Style
Accord Saturn Cavalier
Accord 1 1/3 4
Saturn 3 1 17
Cavalier 1/4 1/7 1
Develop an overall priority ranking
Priorities for each car using each criterion
Price MPG Comfort Style
Accord 0.123 0.087 0.593 0.265
Saturn 0.320 0.274 0.341 0.656
Cavalier 0.557 0.639 0.065 0.080

Priority of each criterion
Price 0.398
MPG 0.085
Comfort 0.218
Style 0.299

Weight each car's priority by the corresponding criterion priority
Accord
0.398(0.123) + 0.085(0.087) + 0.218(0.593) + 0.299(0.265) = 0.265
Saturn
0.398(0.320) + 0.085(0.274) + 0.218(0.341) + 0.299(0.656) = 0.421
Cavalier
0.398(0.557) + 0.085(0.639) + 0.218(0.065) + 0.299(0.080) = 0.314

AHP in Excel

Summary
Approaches for Solving multi criteria decision making
Goal Programming (GP)
Handles multiple criteria situations within the general framework of Linear
Programming
Analytic Hierarchy Process (AHP).
Allows the users to arrive at a prioritized ranking of the decision alternatives.
Summary of the procedure to develop a goal programming model
Step1) dentify the goals and any constraints that reflect resource capabilities or
other restrictions that may prevent achievement of the goal
Step2) Determine the priority level of each goal
Step3) Define the decision variables
Step4) Formulate the constraints in the usual linear programming fashion
Step5) For each goal, develop a goal equation. Deviation variables u
1
+
, u
1
-
are
included in each goal equation.
Step6) Develop the objective function in terms of minimizing a prioritized function of
the decision variables.
Procedure for AHP
Pairwise Comparison,
Make pairwise comparison matrix
Synthesization
Other pairwise comparisons
Develop an overall priority ranking