Job Offers Case Study

Assignment 2 1. Case Study: Job Offers, Chapter 4. Robins decision tree is shown below.
Weights: 0.17 0.33 0.50
Income Rating
100 50.0% 0 Snowfall 58.2 200 70.0% 56.9 400 Madison Publishing TRUE 0 Disposable Income 54.0 100 50.0% 0 Snowfall 49.9 200 70.0% 48.6 400 Decision 54.0 150 MPR Manufacturing FALSE 0 Snowfall 36.0 230 70.0% 35.8 320 Pandemonium PizzaFALSE 50.0 0 50 15.0% 43.3 0 35.8 0 43.3 15.0% 29.2 0 29.2 15.0% 65.3 Job Case 0.35 48.6 0.075 65.3 15.0% 40.3 $1300 0.075 40.3 15.0% 73.6 0.35 56.9 0.075 73.6 15.0% 48.6 $1500 0.075 48.6
Snowfall Rating 25
Magazine Overall Rating Score 56 48.6
75
75 75
50 100
56 56
56.9 73.6
25
25
56
40.3
25 25
50 100
56 56
48.6 65.3
100
37.5
29.2
100 100 0
57.5 80 0
0 0 100
35.8 43.3 50.0
1. The proportional scores must be calculated in the consequence matrix. For example, the snowfall rating when it snows 100 cm is 100(100 cm - 0)/(400 cm - 0) = 25. 2. Annual snowfall obviously is a proxy in this situation. It may not be perfect, depending on other climatic conditions. for example, 200 cm of snowfall in Flagstaff may be very different than 200 cm in Minnesota; snow in Flagstaff typically melts after each storm, but not so in Minnesota. Other possible proxies might be proximity to mountains, average snowpack during February (for example), or the average high temperature during winter months. Another possibility is to consider a non-weather index such as the amount of winter-sports activity in the area. 3. The weights are km = 1/2, ks = 1/3, and ki = 1/6. Calculate these by solving the equations km = 1.5ks, km = 3ki, and km +ks + ki = 1. Using these weights, we can calculate the overall scores at the end of each branch of the tree:
Weights: Ratings:
Madison
MPR Pandemonium
Income 0.17 75 75 75 25 25 25 100 100 100 0
Snowfall Magazine 0.33 0.50 25 56 50 56 100 56 25 56 50 56 100 56 37.5 0 57.5 0 80 0 0 100
Overall Score 49 57 74 41 49 66 29 36 43 50
4. Expected values for the three individual attributes and the overall score are given in the following table: EXPECTED VALUES Pandemonium MPR Madison Income Snowfall $1200 0 cm $1600 231.5 cm $1420 215 cm Magazine Score 95 50 75 Overall Score 50 36 55
5, 6. The cumulative risk profiles are shown below. Note that different alternatives are stochastically dominant for different attributes. Looking at the overall score, MPR is clearly dominated In expected value, Madison Publishing comes in second on each attribute but is a clear winner overall. Likewise, Madison looks good in the overall score risk profile, clearly dominating MPR. A good strategy for Robin might be to go to Madison Publishing in St. Paul and look for a good deal on an apartment. Nevertheless, Robin would probably want to perform some sensitivity analysis on the weights used to determine how robust the choice of Madison is (in terms of the calculated expected values and risk profiles) to variations in those weights.
Legend: Madison MPR Pandemonium
1.00 0.80 0.60 0.40 0.20 0.00 $1,000 $1,200 $1,400 $1,600 Income $1,800
1.00 0.80 0.60 0.40 0.20 0.00 -100 0 100 200 300 400 500 Snowfall
1.00 0.80 0.60 0.40 0.20 0.00 0 20 40 60 80 100 Magazine Score
1.00 0.80 0.60 0.40 0.20 0.00 0 20 40 60 80 100 Overall score
2. Class Problem
COUNTER EXAMPLES AND PROOFS (1) EMV (A) EMV (B) does not always imply that A STOCHASTICALLY DOMINATES B E.g.: consider the following cumulative risk profile of two choices A and B
A
1 0.9
B
Fig.1
EMV (A) = 00 + 51= 5 EMV (B) = 0 0.9 + 10 0.1 = 1 BUT A DOES NOT STOCHASTICALLY DOMINATE B (2) A STOCHASTICALLY DOMINATES B always implies that EMV (A) EMV (B) For nonnegative integers, the expected value can be expressed as:
E [ X ] = (1 F ( x ))dx
0
E.g.: consider the following cumulative risk profile of two choices A and B
F (x)
1
F (x)
By the definition of Stochastic Dominance, v, P( X A v ) P( X B v ) v,1 FA (v ) 1 FB (v )

EMV ( A) EMV (B )
(1 F (v ))dv
0 A
1 FB (v )dv
4.16.
The cumulative risk profile is shown below:

1
0.75 Buy 0.5 Make
0.25
0 30 35 40 45 50 55
Johnson Marketing should make the processor because the cumulative risk profile for Make lies to the left of the cumulative risk profile for Buy. (Recall that the objective
is to minimize cost, and so the leftmost distribution is preferred.) Making the processor stochastically dominates the Buy alternative.
Case Study: Job Offers, Part II 1. From the job offers part I question # 3
Weight for income rating =0.17 Weight for snowfall rating =0.33 Weight for magazine rating =0.50 Using these weights, we can calculate the overall scores at the end of each branch of the tree: Income 0.17 75 75 75 25 25 25 100 100 100 0 Snowfall Magazine 0.33 0.50 25 56 50 56 100 56 25 56 50 56 100 56 37.5 0 57.5 0 80 0 0 100 Overall Score 49 57 74 41 49 66 29 36 43 50
Weights: Ratings:
Madison
MPR Pandemonium
Expected Value of Madison = 8p + 50 Expected Value of Pandemonium = 50 Expected Value of MPR =36 E.g.: EVM(Madison)= [(p){0.15*49+0.7*57+0.15*74}+ (1p){0.15*41+0.7*49+0.15*65}] = 8p+50 The sensitivity analysis gives the following results:
Decision Vs. Prob. Income

59 58 57 56 55 54 53 52
Decision
Expected Overall Score P($1500) Madison 0 50 1 58
MPR 36 36
Pandemonium 50 50
Thus, there is no question about the preference for Madison Publishing; with the probability set at the most pessimistic value of 0, Madison and Pandemonium are equivalent. MPR, of course, is never in the running at all. A one-way sensitivity plot for the probability of disposable income being $1500 for the Madison job generated by Precision Tree is shown above. The one-way plot does not show the alternative policies, only the range of possible outcome results.
2. The two-way sensitivity graph is shown below. The labels indicate the optimal choice in each region. The graph makes good sense! When the weights on snowfall and income are small (and hence the weight for the magazine score is high), Pandemonium is the best, reflecting its strong showing on the magazine. Likewise, when the magazine weight is low, MPR is best, reflecting its strength in income and snowfall, but poor showing on the magazine score.
1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8
ks ki
MPR
Madison Pandemonium
1.0

Job Offers Case Study

Uploaded by

Copyright:

Available Formats

Job Offers Case Study

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Job Offers Case Study

Uploaded by

Copyright:

Available Formats

Assignment 2 1. Case Study: Job Offers, Chapter 4. Robins decision tree is shown below.

Weights: 0.17 0.33 0.50

Magazine Overall Rating Score 56 48.6

35.8 43.3 50.0

Income 0.17 75 75 75 25 25 25 100 100 100 0

Snowfall Magazine 0.33 0.50 25 56 50 56 100 56 25 56 50 56 100 56 37.5 0 57.5 0 80 0 0 100

1.00 0.80 0.60 0.40 0.20 0.00 0 20 40 60 80 100 Magazine Score

1.00 0.80 0.60 0.40 0.20 0.00 0 20 40 60 80 100 Overall score

By the definition of Stochastic Dominance, v, P( X A v ) P( X B v ) v,1 FA (v ) 1 FB (v )

The cumulative risk profile is shown below:

0.75 Buy 0.5 Make

Decision Vs. Prob. Income

Expected Overall Score P($1500) Madison 0 50 1 58

You might also like