Professional Documents
Culture Documents
Duygu Durmuolu
IE-443
HW-3
Part I
1)
2)
3)
VALUE FUNCTIONS
Productivity Enhance Cost Increase Security
x Value x Value
-1 0,00 Low: 0 -2 0,00
0 0,50 High: 150 -1 0,50
1 0,75 Mono: decreasing 0 0,83
2 1,00 Rho: 182,4 1 1,00 SUM
Base: 0,22 0,35 0,43 1
Weights: 0,23 0,37 0,4 1
SCORES (LEVELS)
Stat Quo 0 0 0
HQ/HCost 2 125 0,5
MQ/MCost 1 95 0
LQ/LCost 0,5 65 -1
HQ/HCost
0,70 MQ/MCost
0,60 LQ/LCost
0,50
0 0,2 0,4 0,6 0,8 1
Security Weight
Part II
1)
a)
UD_SELECT THE BEST JOB UD_COMPANY'S PRESTIGE UD_KIND OF CORPORATION UD_COMPANY TYPE
Goal Goal Goal Measure
b)
c)
Assessment summary for Preference Set NEW PREF. SET
Common units
UD_BONUS AMOUNT: A continuous single-measure utility function with the following ranges:
least least most most mid- mid-
preferred preferred preferred preferred preference preference
range Level Utility Level Utility Level Utility
0 200 0.000 1000 1.000 800 0.500
1 1.000
2 0.625
3 0.250
4 0.000
0 0.000
1 0.500
2 1.000
UD_MONTHLY PAYMENT: A continuous single-measure utility function with the following ranges:
least least most most mid- mid-
preferred preferred preferred preferred preference preference
range Level Utility Level Utility Level Utility
0 3000 0.500 5000 1.000 4200 0.750
1 1000 0.000 3000 0.500 2500 0.250
High 1.000
Medium 0.750
Low 0.000
UD_REPUTATION DEGREE: A continuous single-measure utility function with the following ranges:
least least most most mid- mid-
preferred preferred preferred preferred preference preference
range Level Utility Level Utility Level Utility
0 1 0.000 6 1.000 4 0.500
UD_SHIFT DURATION: A continuous single-measure utility function with the following ranges:
least least most most mid- mid-
preferred preferred preferred preferred preference preference
range Level Utility Level Utility Level Utility
0 7 0.000 4 1.000 5.5 0.500
Weights
UD_SELECT THE BEST JOB: a Utility function based on these tradeoffs:
A B
No interactions assessed
Probability Details
d)
1
Utility
0
200 1000
LabelUtility
1 1.000
2 0.625
3 0.250
4 0.000
LabelUtility
0 0.000
1 0.500
2 1.000
Utility
0
1000 5000
y=0.00016x-0.15 (x<=2500)
y=0.0005x-1 (2500< x <=3000
y=0.00021x-0.132 (3000<x<=4200)
y=0.0003125x-0.5625 (4200<x<=5000)
Utility histogram for UD_POSITION DEGREE labels
Label Utility
High 1.000
Medium 0.750
Low 0.000
Label Utility
Utility
0
1 6
Utility
0
4 7
Alternative Utility
UD_Shell-Energy 0.731
UD_Unilever-FMCG 0.580
UD_Yapkredi- Finance 0.401
UD_Delphi Reklam-Media/Ad. 0.367
UD_BNB Danmanlk-Consulting 0.340
For Shell, ease of promotion and company type make different among all indicators. Unilever is at
second place by especially high reputation degree indicator. Yapkredi has average degrees of all
indicators. The last two alternatives have important ratio of position degree indicator.
f)
0.833 UD_Shell-Energy
UD_BNB Danmanlk-Consulting
UD_Yapkredi- Finance
UD_Unilever-FMCG
UD_Delphi Reklam-Media/Ad.
Utility
0.167
0 100
Percent of Weight on UD_WORKLOAD Goal
According to graph, workload is not sensitive due to no change in top ranking at sensitive
point.
0.875 UD_Shell-Energy
UD_Delphi Reklam-Media/Ad.
UD_Unilever-FMCG
UD_BNB Danmanlk-Consulting
UD_Yapkredi- Finance
Utility
0.240
0 100
Percent of Weight on UD_STATUS Goal
1.000 UD_Unilever-FMCG
UD_Shell-Energy
UD_Yapkredi- Finance
UD_Delphi Reklam-Media/Ad.
UD_BNB Danmanlk-Consulting
Utility
0.000
0 100
Percent of Weight on UD_COMPANY'S PRESTIGE Goal
Utility
0.306
0 100
Percent of Weight on UD_SALARY Goal
g)
1.000
Utility
0.000
UD_SELECT T HE BEST JOB UD_SALARY UD_WORKLOAD UD_ST AT US UD_COMPANY'S PREST IGE
In accordance with graph, overtime rating and ease of promotion indicators provides the
supremacy of Shell over Unilever. Unilevers better indicator reputation degree cannot give
enough utility to be ranked as 1st
2)
Ranking for UD_SELECT THE BEST JOB Goal
Alternative Utility
UD_Unilever-FMCG 0.982
UD_Shell-Energy 0.731
UD_Yapkredi- Finance 0.436
UD_Delphi Reklam-Media/Ad. 0.367
UD_BNB Danmanlk-Consulting 0.362
Alternative Utility
UD_Shell-Energy 0.651
UD_Unilever-FMCG 0.518
UD_Delphi Reklam-Media/Ad. 0.406
UD_Yapkredi- Finance 0.364
UD_BNB Danmanlk-Consulting 0.312
We added new objective, risk and its measure, percentage of risk of lower position. We
assigned high percentage of risk of lower position to Yapkredi, therefore ranking results are
changed. Yapkredi became 4th and Delphi became 3rd.
Part III
1)
a)
b)
C)
Assessment summary for Preference Set NEW PREF. SET
Common units
UD_COMPANY'S PRESTIGE: a Utility function assessed using AHP:
Weights
UD_SELECTING THE BEST JOB: a Utility function assessed using AHP:
No interactions assessed
Probability Details
d)
Ranking for UD_SELECTING THE BEST JOB Goal
Alternative Utility
UD_UNILEVER 0.363
UD_SHELL 0.255
UD_YAPIKREDI 0.160
UD_DELPHI 0.158
UD_BNB 0.063
e)
0.469 UD_UNILEVER
UD_SHELL
UD_YAPIKREDI
UD_DELPHI
UD_BNB
Utility
0.034
0 100
Percent of Weight on UD_COMPANY'S PRESTIGE Measure
According to graph, companys prestige graph is not sensitive due to no change in top
ranking at sensitive point.
0.444 UD_DELPHI
UD_UNILEVER
UD_SHELL
UD_YAPIKREDI
UD_BNB
Utility
0.052
0 100
Percent of Weight on UD_SALARY Measure
0.508 UD_YAPIKREDI
UD_UNILEVER
UD_SHELL
UD_DELPHI
UD_BNB
Utility
0.036
0 100
Percent of Weight on UD_STATUS Measure
Utility
0.039
0 100
Percent of Weight on UD_WORKLOAD Measure
f)
1.000
Utility
0.000
UD_SELECTING THE BEST JOB UD_COMPANY'S PRESTIGE UD_SALARY UD_STATUS UD_WORKLOAD
In accordance with graph, monthly payment and reputation degree indicators provides the
supremacy of Unilever over Shell. Shells better indicators overtime and ease of promotion
cannot give enough utility to be ranked as 1st
Part IV
Our rankings changed when AHP method was used instead of SUF method. The reason might
be that in SUF we compared measures and tried to rank them but in this method we
compared objectives in a pair-wise way so that we assigned relative importance level by
considering each and every pair. In addition, in AHP each and every pair of alternative were
compared and relative importance values were assigned, based on each objective.
Moreover, in SUF, weights are decided by fitting values into functions but in AHP, values are
summed for each objective in a column-wise way and then normalized in order to find
weights. All these differences between methods might cause differences in ranking of
alternatives.