DECISION TREES

Dr. Vaidy Jayaraman

Caselet # 1
Political Supply Chain Systems (PSCS) is a newly formed computer service firm that is in the
process of selecting a new computer system. For PSCS, the final decision will be to lease one of
three computer systems which differ in size and capacity. The three decision alternatives are:

d1 – Lease a large computer system

d2 – Lease a medium sized computer system
d3 – Lease a small computer system.

For PSCS, the states of nature are: s1 – High customer acceptance; s2 – Low customer
acceptance of PSCS services.

PAYOFF MATRIX
S1 S2
d1 200000 -20000
d2 150000 20000
d3 100000 60000

Given P(S1) = 0.3; P(S2) = 0.7 find

a) Find EVWI
b) Find EVPI

Suppose that PSCS decides to consider hiring a market research firm to study the potential
acceptance of PSCS services. The outcomes of PSCS marketing research study are as follows:
I1 – Favorable market research report
I2 – Unfavorable market research report
In PSCS case, the past record of the marketing research company on similar studies has led to
the following estimates of relevant conditional probabilities:
I1 I2
s1 0.8 0.2
s2 0.1 0.9

Calculate EVII.
Caselet # 2

L&T is looking into the possibility of expanding their operations in Columbia and
has invited a contractor to bid on a construction job. The value of the contract
depends on the length of time it takes to complete the project. If the project is
finished on time, there is a profit of $50,000. IF the contractor is late finishing the
project, he will lose $10,000. Weather is the SOLE determinant of whether the
project will be late. If the weather is good, the project will be completed on time;
if it is bad, the project will not be completed on schedule. Based on his past
experience the contractor’s subjective probability of good weather is 20%. The
contractor, however, has the opportunity to buy a long-range forecast from an
independent weather-forecasting company. The weather-forecasting company
has a fairly good track record for these long-range forecasts. Its files indicate that
70% of the time it successfully predicted good weather, and 80% of the time it
was able to predict bad weather. In other words,

If I1 = prediction of good weather

I2 = prediction of bad weather
S1 = good weather
S2 = bad weather

P(I1/S1) = 0.7 P(I1/S2) = 0.2

P(I2/S1) = 0.3 P(I2/S2) = 0.8

The cost of weather-forecasting service is $5,000

Caselet # 3
A manufacturer of electronic accessories for use in computers currently has a
capacity of 40000 pieces per month. The business strategy group for the company
recently performed a forecasting exercise to assess the emerging demand for the
accessories in the next five years. The study revealed that there is a 40%
probability that the growth in demand for the accessories will be strong during
this planning horizon and a 60% probability that the growth in demand will be
moderate. The study identified three options fir the manufacturer to augment
capacity.
Option 1 is to expand the capacity by adding new capacity to meet the demand.
Option 2 is to augment capacity in the existing factory itself by some de-
bottlenecking operation (in which case there will be limits to capacity expansion).
Option 3 is to go for subcontracting. The sub-contracting option adds marginal
capacity and could be used as a temporary arrangement. If there is a strong
growth in demand, then the new capacity could be added a year later. The study
revealed the following additional information about the merging scenario and the
costs and benefits of each of the alternatives:

 The cost of adding new capacity is Rs 750000. This cost goes up by 5% if it is

deferred by a year.
 The cost of expanding in the exiting factory itself is Rs. 275000
 The cost of subcontracting is negligible, as no major investments are
envisaged either at the supplier side or at the manufacturer side.
 The revenue accruing from new capacity is as follows: if it is a strong
growth, the revenue will be Rs. 850000 and in the case of a moderate
growth, the revenue will be Rs. 400000. These figures do not change even if
the new unit comes into operation a year later.
 The revenue accruing from the expansion of the existing capacity is as
follows: if it is a strong growth, the revenue will be Rs. 550000 and in the
case of a moderate growth, the revenue will be Rs. 300000.
 The revenue accruing from the existing factory with subcontracting is as
follows: if there is strong growth, the revenue will be Rs. 350000 and in the
case of moderate growth, the revenue will be Rs. 180000.

Arrive at an appropriate capacity planning strategy using a decision tree.

(All revenues are yearly figures).
Caselet # 4

A builder has located a piece of property that she would like to buy and
eventually build on. The land is currently zones for four homes per acre, but she is
planning to request new zoning. What she builds depends on approval of zoning
requests and your analysis if this problem to advise her. With her input and your
help, the decision process has been reduced to the following costs, alternatives
and probabilities:

Cost of land: $2 million

Probability of rezoning: 60

If the land is rezoned, the contractor must decide whether to build a shopping
center or 1500 apartments that the tentative plan shows would be possible. If she
builds a shopping center, there is a 70% chance that she can sell the shopping
center to a large department store chain for $4 million over her construction cost,
which excludes the land; and there is a 30% chance that she can sell it to an
insurance company for $5 million over her construction cost (also excluding the
land). If, instead of the shopping center, she decides to build the 1500
apartments, she places probabilities on the profits as follows: there is a 60%
chance that she can sell the apartments to a real estate investment corporation
for $3000 each over her construction cost; there is a 40 percent chance that she
can get only $ 2000 each over her construction cost. (both exclude the land cost).

If the land is not rezoned, she will comply with the existing zoning restrictions and
simply build 600 homes, on which she expects to make $ 4000 over the
construction cost on each one (excluding the cost of land).

Draw a decision tree of the problem and determine the nest solution and the
expected net profit.
Caselet # 5

A retailer must decide whether to build small or a large facility at a new location.
Demand at the location can be either low or high, with probabilities estimated to
be 0.4 and 0.6 respectively. If a small facility is built and demand proves to be
high, the manger may choose not to expand (payoff = $223000) or to expand
(payoff=$270000). If a small facility is built and demand is low there is no reason
to expand and the payoff is $200000. If a large facility is built and demand proves
to be low, the choice is to do nothing ($40000) or to stimulate demand through
local advertising. The response to advertising may be either modest or sizable
with their probabilities estimated to be 0.3 and 0.7, respectively. If it is modest,
the payoff is estimated to be only $ 20000; the payoff grows to $220000 if the
response is sizable. Finally, if a large facility is built and demand turns out to be
high, the payoff is $800000.

Draw a decision tree. Then analyze it to determine the expected payoff for each
decision and event node. Which alternative-building a small facility or building a
large facility- has the higher expected payoff?