Transcript Do National Marketing

DECISION ANALYSIS
Decision Analysis
•
•
•
•
•
•
•
Introduction
Problem Formulation
Decision Making without probabilities
Decision making with probabilities
Risk analysis and sensitivity analysis
Decision analysis with sample information
Computing Branch probabilties
Decision Analysis
Decision analysis is used to develop strategy when
decision maker is faced with several decision
alternatives nad a risk filled pattern of future events.
Problem Formulation: Verbal statement of the problem.
Identification of the alternatives, the uncertain future
events and the consequences associated with each
decision alternative.
Example: Investment of a 1.000.000,00 Euro.
Buy a house
Stock Market
Buy a shopping area
Start a new business
Influence Diagrams
A graphical device that shows the relationships among the decision, the
chance events and the consequences for a decision problem. The node used
to represent the decision, Rectangles or squares used to represent decision
nodes, circles or ovals are used to depict chance nodes and diamonds are
used to depict consequences. Lnes connect the nodes show the direction of
influence.
State of Nature
Strong (s1)
Demand
Complex
Weak (s2)
Profit
Consequences
Decision Alternatives
Small Complex (d1)
Medium complex (d2)
Large complex (d3)
Profit
Payoff Tables
Payoff tables: Consequences of the alternative decisions.
State or nature.
Used in mono criterion approaches: To maximize the
profit.
Stater of Nature
Decision Alternatives Strong demand Weak Demand
Small Complex
Medium Complex
Large Complex
8
14
20
2
5
-9
Decision Trees
A graphical presentation of the decision making process
Strong (s1)
8
Small (d1)
2
Weak (s2)
2
Strong (s1)
14
Medium (d2)
1
3
Weak (s2)
5
Strong (s1)
20
Large (d3)
4
Weak (s2)
-9
Decision Making without probabilities
Not required knowledge of the probabilities of the state of
nature.
Optimistic Approach: Evaluates decision alternatives of
the best payoff that can occur.We select alternative with
the best overall maximum payoff.
Conservative Approach: Evaluates decision alternatives of
the worst payoff that can occur.We select alternative
with the best overall minimum payoff.
Decision Alternatives Maximum Payoff
Decision Alternatives Minimum Payoff
Small Complex
Medium Complex
Large Complex
Small Complex
Medium Complex
Large Complex
8
14
20
7
5
-9
Minimax Regret Approach
Minimax regret approach (neither optimistic nor purely
conservative).
Opportunity loss - regret: Difference between the payoff for the best
decision alternative and the payoff of the conservative one.
Decision Alternatives
Small Complex
Medium Complex
Large Complex
Decision Alternatives
Small Complex
Medium Complex
Large Complex
Decision Alternatives Strong demand-Regret
Strong demand
8
14
20
Small Complex
Medium Complex
Large Complex
Best Alternative
Weak demand
7
5
-9
Decision Alternatives Weak demand-Regret
Best Alternative
Small Complex
Medium Complex
Large Complex
Decision Alternatives Maximum Regret
Small Complex
Medium Complex
Large Complex
12
6
0
12
6
16
Minimum regret
0
2
16
Pay off Table
A newspaper vendor purchases the paper at $0.15 and
sells them for $0.50. The demand for papers is as follows:
Demand
Probability
0
0.05
1
0.10
2
0.20
3
0.30
4
0.20
5
0.15
The vendor is interested in determining how many
newspapers should be purchased in order to maximize
expected profit.
Analysis
. The minimum number of newspapers to be purchased is
0.
. The maximum number of newspapers to be purchased
is…?
. Evaluation: Determine what happens when the vendor
purchases k newspapers.
Pay off Table
Probability Demand
0,05
0,1
0,2
0,3
0,2
0,15
Purchase Prise
Seling Prise
0
1
2
3
4
5
0
0
0
0
0
0
0
0,15
0,5
Number of Newspapers Purchased
1
2
3
4
-0,15
-0,3
-0,45
-0,6
0,35
0,2
0,05
-0,1
0,35
0,7
0,55
0,4
0,35
0,7
1,05
0,9
0,35
0,7
1,05
1,4
0,35
0,7
1,05
1,4
Marginal profit
Marginal Loss
0,35
0,15
5
-0,75
-0,25
0,25
0,75
1,25
1,75
Expected Profit
Probability Demand
0,05
0
0,1
1
0,2
2
0,3
3
0,2
4
0,15
5
Expected Profit
Purchase Prise
Seling Prise
0
0
0
0
0
0
0
Number of Newspapers Purchased
1
2
3
4
-0,15
-0,3
-0,45
-0,6
0,35
0,2
0,05
-0,1
0,35
0,7
0,55
0,4
0,35
0,7
1,05
0,9
0,35
0,7
1,05
1,4
0,35
0,7
1,05
1,4
0 0,325
0,15
0,5
0,6 0,775
Marginal profit
Marginal Loss
5
-0,75
-0,25
0,25
0,75
1,25
1,75
0,8 0,725
0,35
0,15
How Many to buy?
. The previous table shows that if the vendor purchases 4
newspapers, then on average, the vendor would make
$0.80 in profit. All others have lower expected values.
Hence, the vendor should purchase 4 newspapers.
. Note that if the vendor purchases 4 newspapers, then on
any given day, the vendor would make {-$0.60, -$0.10, $0.40, $0.90, or $1.40}, not $0.80. However, in the long
run, the vendor would average $0.80.
. Good homework assignment: Set up the Excel table so
that you can make changes to the parameters of the
problem (probabilities and prices)and the table would
automatically reflect the changes.
SciTools Inc
Probability of No Bids =0,30
Probability of a Bid = (1-0,30)=0,70
Given a bid, the probability that the bid will be:
Low Bid
Probability
Less than 115K
0,20
Between 115K amd 120 K
0,40
Between 120K and 125K
0,30
Greater than 125K
0,10
Event
No Bid
Probability
0,30
Bid Less than 115K
(0,70 X 0,20)
0,14
Bid between 115K and 120K
(0,70X0,4)
0,28
Bid between 120K and 125K
(0,70 X 0,30)
0,21
Bid greater than 125K
(0,7 X 0,10)
0,07
Create Pay-off Table
Cost of Bid = 5.000,00
Profit = (Value of Bid) – 95.000,00
If we bid 115K and win, then our profit will be 115 K –
95K-5K = 15K
If we bid 115K and lose, then our loss will be –5K
Competitors bid
Our Bid <115K 115 .. 120 120 … 125 >125 No Bid Expected Value
No Bid
0
0
0
0
0
0
115
-5
15
15 15 15
12,2
120
-5
-5
20 20 20
9,5
125
-5
-5
-5 25 25
6,1
Probability 0,14 0,28
0,21 0,07 0,3
Case Study
New product in the Market..
A) Test the market
B) Do a National Marketing
Available Information
- Fixed cost for Test Marketing 3000
- Fixed Cost for National Marketing:90000
- The following joint probabilities coming as a result of the
conditional probabilities
National Market
Test Market
Test Great
Test Fair
Test Awful
Great
Fair
0,24
0,18
0,005
0,425
0,045
0,3
0,025
0,37
Awful
0,015
0,12
0,07
0,205
Thus the probabilities that the national market will be Great, Fair, or Awful are 0,425 0,370 and
0,205 respectively.
No
0,42,
Nat.
marketing
No
-90000
Great
0,37
Yes
Fair
0,205
ACME
TEST Market
Awful
No
Great
0,80,
Nat.
marketing
0,15
Yes
-90000
0,3
Great
Fair
0,05
Awful
No
Yes
0,6
Fair
Great
Nat.
marketing
-90000
0,30
Yes
0,50,
0,20
0,1
Fair
Awful
No
Awfull
0,05
Nat.
marketing
Yes
0,25
Great
Fair
-90000
0,70
Awful
Conduct a Test
• Outcomes
Test marketing is Great (Probability = 0,30)
Choices
Do a National Market
Do not do a National Market
Test Marketing is Fair (Probability = 0,60)
Choices
Do a National Market
Do not do a National Market
Test Marketing is Awful (Probability = 0,10)
Choices
Do a National Market
Do not do a National Market
Prise: 18
Demand – Grate: 60000, Fail: 30000, Awful: 600
Do a Direct National Marketing
Prise
Cost
OutComes
Great
Fair
Awful
18
Demand
6000
3000
900
Expected Payoff
Cost
Net
Payoff
108000
54000
16200
69201
90000
-20799
Probabilit
y
0,425
0,37
0,205
Results 1
Thus, if we go for a direct National marketing campaign,
it is going to result in a expected loss of 20,799.00
Since this is a loss, we do not want to go this approach
Let us now consider whether doing a test marketing will
lead to better outcomes
Test market
Probability
0,6
0,3
0,1
What do we
Outcomes
do?
Payoff
Great
?
?
Fair
?
?
Awful
?
?
Test Market is Great
Do not do a National market
- Payoff
200 units @ 18.00 = 3600
Cost
= -3000
Net
= 600
Do the National Marketing
Great
Demand
Payoff
Probability
Fair
Awful
6000
3000
900
108000
54000
16200
0,8
0,15
0,05
Expected Payoff
95310
National Marketing Cost
-90000
Payoff from Test marketing
Cost of Test Marketing
Net Profit
3600
-3000
5910
What should we do?
If the test market says “Great”. We have two choices.
- do not do a National Marketing, in which case, we and
up with a profit of 600
- Do a National Marketing, in which case we end up with
a profit of 5910
- Hence, if we test market result is great then do a
National marketing with a payoff of 5910
Test Market - Fair
Do not do a National market
- Payoff
100 units @ 18.00 = 1800
Cost
= -3000
Net
= -1200
Great
Demand
Payoff
Probability
Fair
Awful
6000
3000
900
108000
54000
16200
0,3
0,5
0,2
Expected Payoff
National Marketing Cost
Payoff from Test marketing
Cost of Test Marketing
Net Profit
62640
-90000
1000
-3000
-29360
What should we do?
If the test marketing is Fair.
- If we do not do a National Marketing, we end up with
a loss of 1,200.00
- If we do a National Marketing, we end up with a loss
of 28,580.00
- Pick the lesser of the losses (-1,200).
- Do not do a National Marketing if the test Marketing is
Fair
Test Market - Awful
Do not do a National market
- Payoff
30 units @ 18.00 = 540
Cost
= -3000
Net
= -2400
Great
Demand
Payoff
Probability
Fair
Awful
6000
3000
900
108000
54000
16200
0,05
0,25
0,7
Expected Payoff
National Marketing Cost
Payoff from Test marketing
Cost of Test Marketing
Net Profit
30240
-90000
540
-3000
-62220
What do we do?
Choices is Fair.
- If we do not do a National Marketing, we end up with
a loss of 2,460.00
- If we do a National Marketing, we end up with a loss
of 62220,00
- Neither is good … Pick the lesser of the losses.
- Do not do a National Marketing if the test Marketing is
Awful
If we do a test
Test market
Probability
Outcomes
0,3
Great
0,6
Fair
0,1
Awful
OVERALL
What do we
do?
Payoff
Do a
National
Marketing
5910
Do not do a
National
Marketing
-1200
Do not do a
National
Marketing
-2460
807
Choices
Do a Test
- test results Great
- Do not do national marketing (Profit of 600)
- Do National Marketing (Profit of 5910)
- Test results Fair
- Do not do national marketing (loss of 1200)
- Do National Marketing (loss of 28500)
- Test results Awful
- Do not do national marketing (0)
- Do National Marketing (loss of 20799)
Overall
Do a market test
If the test is great then do a National Marketing
If the test results is fair then do not do a National
Marketing
If the test is Awful then do not do a National Marketing
Expected Value of Information
We have three possible information states in the ACME
example.
• No information (No test marketing)
• Sample information (Test marketing)
• Perfect information
Perfect Information
What if we are able to predict with absolute certainty,
whether the product will be great, Fair or Awful? What
would our decision be?
Market is Great
If we market the product:
Demand = 6000
Revenue = 6000 * 18 =108000
Cost = 90000
Net profit = 18.000
So, if we knew that the market is going to be great we will
market the product and make 18.000 in profit
Market if Fair
If we market the product
Demand = 3000
Revenue = 3000 * 18 = 54000
Cost = 90000
Net Loss = (38000)
Since marketing the product would result in a loss, we
would not market the product if we knew with
certainty that the market of Fair
Market is Awful
In this case, marketing the product would result in a loss
of (73800). So we would not market the product.
With Perfect Information
Outcome
Payoff
Probability
Great
18000
0,425
Fair
0
0,37
Awful
0
0,205
Expected Profit
7650