MT 231: QUANTITATIVE METHODS I
Zeymisco Elias: Mob: 0763266714
SAUT-MWANZA
November 19, 2024
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
MT 231: QUANTITATIVE METHODS I
Zeymisco Elias: Mob: 0763266714
SAUT-MWANZA
November 19, 2024
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
The value of perfect and imperfect information.
• Perfect information is available when a 100% accurate
prediction can be made about the future.
• Imperfect information: The forecast is always correct but
can be incorrect. Imperfect information is not as useful as
perfect information.
• The value of information (Perfect or imperfect)= Expected
value with (Perfect or imperfect) information - Expected value
without (Perfect or imperfect) information.
Notation: V OI(P ) = EOP − EW P or
V OI(IM ) = EOIM − EW IM .
• The value of information for perfect or imperfect
information are calculated in the same manner.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Example 1
Consider the following data on the projects that a person
wants to take;
Demand POOR OK STRONG
Probability 30% 50% 20%
Project A 100 140 200
Project B 90 110 210
Calculate the value of perfect information.
Solution
V OI(P ) = EOP − EW P .
Expected value with (Perfect or imperfect) information
(VOI(P)) = 30% · 100 + 50% · 140 + 20%210 = 142.
EV (A) = 30% · 100 + 50% · 140 + 20% · 200 = 140.
EV (B) = 30% · 90 + 50% · 110 + 20% · 210 = 124. Since
EV (A) > EV (B) then Expected value without (Perfect or
imperfect) information (EWP)=EV(A)=140. So
V OI(P ) = EOP − EW P = 142 − 140 = 2.
Therefore the value of perfect information is 2.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Example 2
A man runs a cafeteria that is located at a city center. He
decides to collect information on the demand and supply of
food to customers around the cafeteria. The number of
customer is uncertain. He has analyzed daily sales over the
past six months and establish four possible demand levels and
their associated probabilities as portrayed in the table below;
What is the maximum amount that a man is willing to pay for
this information to the nearest whole $?.
Solution
Expected value with Perfect information (V OI(P )) =
0.15 × 1170 + 0.3 × 1612 + 0.4 × 2015 + 0.15 × 2496 = 1839.5$.
EV (450) = 1170$
EV (620) = 1517.2$
EV (775) = 1648.25$
EV (960) = 1586.4$
Since the largest expected value is EV (775) = 1648.25$ then
Expected value without Perfect information
=EV (775) = 1648.25$.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Example 2 Cont..
So
V OI(P ) = EOP − EW P = 1839.5$ − 1648.25$ = 191.25$.
Therefore the amount that a man will be willing to pay is
191.25$.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Marginal Probability: Maximax and Minimax Rules.
• As individuals we need to make decisions in our daily life.
Decision can be made from alternatives that we see in our
surrounding eg. what type of vehicle to buy, purchase a plot,
investing in stocks etc.
• The common decision making rules are:-
i Decision making under certainty.
ii Decision making under uncertainty.
iii Decision making under risk.
• Decision making under uncertainty are those problems in
which each course of action can result in alternative outcomes
but the probabilities of these outcomes cannot be assigned as
a decision maker does not know which outcome can or will
occur.
eg a farmer may know the possible outcomes for the monsoon
of this year as heavy, moderate or low rainfall but does not
know the probability of its occurrence as the farmer may not
have the past information in order to make predictions.
• Decision making under risk, the decision maker has
additional information in terms of probabilities or chances of
having heavy, moderate or low rainfall based on past
knowledge and records.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Decision Making under uncertainty
• With reference to the previous example, a farmer would
know that the yield of crop will be high, low, or moderate but
will not have enough information to be able to use probability
for the possible state that will occur.
Suppose the farmer wants to cultivate crop A, crop B, and
crop C and he knows that the possible yield are high,
moderate, or low with the following payoff (Profits in 1000’s
Tsh) as shown in the table below;-
Table: 1
States of nature
Alternative strategies High Moderate Low
Cop A 60 42 -10
Crop B 85 60 -20
Crop C 50 25 -12.5
The table 1 is called the payoff table.
• Now we will try to help the farmer in decision making by
demonstrating the following decision rules;-
a The maximax criterion
b The minimax regret criterion
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
The Maximax criterion
• The maximax criterion is based on optimistic rule assuming
that the best situation will occur in the future.
Now, we prepare the payoff matrix;
Table: The payoff matrix in Tsh 1000’s
Crop A 60 42 -10
Crop B 85 60 -20
Crop C 50 25 -12.5
• Select the maximum payoff for each alternative:
Table: The maximum payoff in Tsh 1000’s
Crop Payoff
Crop A 60
Crop B 85
Crop C 50
• Select the highest payoff from these payoffs.
The highest payoff is 85000 from crop B. Therefore crop B
will be recommended.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
The Minimax regret criterion
• The technique calculate the opportunity loss or regret for
not taking the best decision under each state of nature.
Table: The payoff matrix in Tsh 1000’s
Crop A 60 42 -10
Crop B 85 60 -20
Crop C 50 25 -12.5
• Prepare regret payoff matrix.
Table: 5. Regret payoff matrix
Regret payoff Maximum Re
Alternative strategies High Moderate Low
Cop A 25 18 0 25
Crop B 0 0 10 10
Crop C 35 35 2.5 35
From 5 the minimum regret is observed for crop B (
regret=10), therefore crop B will be recommended.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Opportunity Loss
• Opportunity loss is the difference between an actual payoff
for a decision and the optimal payoff for that state of nature.
Table: Opportunity Loss Table
Act
State of nature A1 A2 A3
S1 M1 − P11 M1 − P 12 M1 − P13
S2 M2 − P21 M2 − P22 M2 − P23
S3 M3 − P31 M3 − P32 M3 − P33
Where Mi (1, 2, 3)= Maximum payoff for each state of nature.
Pij =Actual payoff of each state.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Example
Consider the following payoff table;-
Profit in $ 1000’s
Investment choice Strong Stable economy Weak
Large factory 200 50 -120
Average factory 90 120 -30
Small factory 40 30 20
What will be the opportunity loss for each act.
Solution
Prepare opportunity loss (regret) table.
Opportunity Loss in $ 1000’s
Investment choice Strong Stable economy Weak
Large factory 0 70 140
Average factory 110 0 50
Small factory 160 90 0
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Decision tree
• A tree is a sequence of branches and subbranches.
• It is technique for solving complex problems.
• If different branches and subbranches of a tree are
associated with course of action, the states of nature,
probabilities of states of nature and payoffs of a given decision
making problem such a tree is know a decision tree.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Parts of a decision tree
1 Decision point
A point in the pictorial representation of a decision tree
having a course of action as immediate subbranches.
A square ( ) is used to represent a decision point.
2 Chance point|Chance node| event point
A point in the pictorial representation of a decision tree
having states of nature as immediate subbranches. A
circle ( ) is used to represent a chance point.
3 Root node
The direction of a tree is from left to right, the leftmost
node is known as a root node.
Represented by a square as a decision point.
4 Decision and chance branches
Branches that represent course of actions are called
decision branches while branches that represent states of
nature are called chance branches.
The probabilities corresponding to states of nature are
written along the chance branches.
5 End point
The points in the pictorial representation of a decision
tree which are neither followed by any direction branch
nor by chance branches.
6 Terminal branch
Any branch that ends at an end point.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Decision tree cont..
Figure: Pictorial representation of a typical decision tree
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Example 1
Draw a decision tree from the following payoff table;-
Profit in $ 1000’s (Nature of states)
Investment choice Strong Stable economy Weak
Large factory 200 50 -120
Average factory 90 120 -30
Small factory 40 30 20
Probability 0.3 0.5 0.2
Solution
Figure: A decision tree
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
Example 2
Consider the following payoff table.
Alternatives Growing Declining
Stocks 70 -13
Mutual funds 53 -5
Bonds 20 20
Probability 0.4 0.6
Solution
Figure: A decision tree
Since the maximum EV = 20.2 from stocks then investing in
stocks is the best choice.
Zeymisco Elias: Mob: 0763266714 MT 231: QUANTITATIVE METHODS I
