DSME 6620

Decision Models and Applications

Lecture Eight

Advanced Topics
in Mathematical Programming II

1
DSME 6620
Decision Models and Applications

Agenda

• Optimization in presence of private data

– Product line design problem

• Multi-stage decision making

– Aggregate planning problem

2
DSME 6620
Decision Models and Applications

Optimization in Presence of
Data Privacy

• Another type of uncertainty in the decision making

is caused by the data privacy. That is, some pieces of
data/information are privately owned/known by only
some parties and thus are not available to others.

• Decision makers should take into account data

privacy in problem formulation and solving.

• On the other hand, proper mechanisms can be

devised to screen or signal the private data.

3
DSME 6620
Decision Models and Applications

Product Line Design Problem

Suppose a firm that can design and make a line of

products differentiated along a quality dimension. The
firm enters into a new market with a single consumer,
whose valuation for quality is a piece of private data and
hence unknown to the firm. Specifically, the valuation
depends on whether the consumer type is high-end or low-
end.

1) If the consumer is of the high-end type, by purchasing

a product with quality level q, he/she gains a utility of
10q;
2) If the consumer is of the low-end type, by purchasing
a product with quality level q, he/she gains a utility of
6q.

The consumer buys one unit or nothing at all. There are

constant marginal costs of production at any given quality
level, and the marginal
cost of producing a product with quality q is q2.

4
DSME 6620
Decision Models and Applications

Product Line Design Problem

The firm has a prior belief about the type of the consumer:
there is a half-half chance the customer is of high-end or
low-end.

For any product designed for the market, the firm needs to
decide on its quality q and its selling price p. A consumer
is willing to choose a product design only if the net utility
gained by the consumer (which is the utility minus the
price) is nonnegative. If multiple options are available in
the market, the consumer will choose the one that yields
the highest (nonnegative) net utility.

For the firm, the purpose of decision making is two-fold.

First, the firm aims to maximize the profit collected from
selling to the consumer, while ensuring the customer’s
participation. Second, in this process, the firm wants to
learn the private data associate with consumer, i.e., his/her
actual type.

5
DSME 6620
Decision Models and Applications

Product Line Design Problem

Analysis & Formulation

What happens if the firm ignores the second purpose? In
other words, there was no need to learn the private data
associated with consumer?

6
DSME 6620
Decision Models and Applications

Product Line Design Problem

Analysis & Formulation

7
DSME 6620
Decision Models and Applications

Multi-Stage Decision Making

• In reality, decision making is not necessarily of one

shot only. Instead, we may want to make decisions
sequentially, i.e., in several stages/phases/periods.
Hence, multi-stage decision making.

• If all the information of future stages is known and

deterministic, then, mathematical programming can
be a viable approach to formulating and solving the
multi-stage problem.

• Otherwise, we need other decision-making tools,

such as decision analysis, Markovian decision
process/ dynamic programming, stochastic network
programming, business process simulation, etc.
– In particular, as will be discussed next, decision
analysis can be used to model multi-stage problems
with uncertainty.

8
DSME 6620
Decision Models and Applications

Aggregate Planning Problem

During the next 4 months, the SureStep Company must

meet (on time) the following demands for pairs of
shoes:

Month Demand of shoes

(pairs)
1 3,000
2 5,000
3 2,000
4 1,000

At the beginning of the first month, 1500 pairs of

shoes are on hand, and SureStep has 100 workers.

9
DSME 6620
Decision Models and Applications

Aggregate Planning Problem

A worker is paid $1500 per month. Each worker can

work up to 160 hours a month before he or she receives
overtime. A worker can work up to 20 hours of overtime
per month and is paid $13 per hour for overtime labor.

It takes 4 hours of labor and $15 of raw material to

produce a pair of shoes.

At the beginning of each month, workers can be hired or

fired. Each hired worker costs $1600, and each fired
worker costs $2000. At the beginning of each month, a
holding cost of $3 per pair of shoes left in inventory is
incurred. All of the production in a given month can be
used to meet that month’s demand.

SureStep wants to use LP to determine its optimal

production schedule and labor policy.

10
DSME 6620
Decision Models and Applications

Aggregate Planning Problem

Analysis & Formulation

11
DSME 6620
Decision Models and Applications

Aggregate Planning Problem

Analysis & Formulation

12