Title: TM%20792%20Special%20Topics%20Decision%20Theory%20June%202,%202008%20%20Spring%202008%20Dr.%20Frank%20Joseph%20Matejcik
1TM 792 Special Topics Decision TheoryJune 2,
2008 Spring 2008Dr. Frank Joseph Matejcik
Ch 7 cont. problems Ch 8 Revising by New
Information Ch 9 Biases in Probability Assessment
2Agenda New Assignment
- Tentative Schedule
- Assignment Ch 7 1, 3 Ch 8 3, 8
- Chapters 7 problems, 8, and start of 9 GW
- We will do the survey online
- Decision Analysis for Management Judgment 3rd
Edition Paul Goodwin George Wright, John Wiley
EU - Many slides solutions provided by John Wiley.
3(No Transcript)
4 Access Overview
- Instructor Dr. Frank J. Matejcik CM 319
- Work (605) 394-6066 roughly 10-3 M-F in May
- Cell (605) 431-5731 Ill try to keep it
nearby - Home (605) 342-6871 Call in June?
- Frank.Matejcik_at_.sdsmt.edu
- Do the book, mostly
5Web Resources
- Class Web site on the HPCnet system
- http//sdmines.sdsmt.edu/sdsmt/directory/courses/2
008su/tm792M081 - www.wileyeurope.com/go/goodwinwright/
- Streaming video ? http//its.sdsmt.edu/Distance/
- The same class session that is on the DVD is on
the stream in lower quality. http//www.flashget.c
om/ will allow you to capture the stream more
readily and review the lecture, anywhere you can
get your computer to run.
6Tentative Schedule
Chapters Assigned 12-May 1,2,3 e-mail,
contact Discussion Q. 3 page 25 19-May 3,4,5
Ch 3 1, 3bc, 6 Ch 4 5, 10 26-May
Holiday 29-May 5,6,7 Ch 5 3, 8 Ch 6 4, 8
Holiday makeup 02-June 7,8,9 Ch 7 1, 3 Ch 8
3, 8 09-June 9,10 16-June 11,12 23-June
13,14 30-June Overview, Final
Attendance Policy Help me work with you.
7Chapter 7 problem 2
- The managers of a food company are about to
install a number of automatic vending machines at
various locations in a major city. A number of
types of machine are available and the managers
would like to choose the design which will
minimize the profit that will be lost because the
machine is out of order. The following model is
to be used to represent the lost profit - Cost of lost profit per month
- (number of breakdowns per month)
- x (time to repair machine after each
breakdown, in hours) - x (profit lost per hour)
- One machine that is being considered is the
Supervend, and the following probability
distributions have been estimated for this
machine
8 Chapter 7 problem 2
- Number of Repair Average
- breakdowns time profit lost
- per month Prob. (hours) Prob. per hour Prob.
- 0 0.5 1 0.45 10 0.7
- 1 0.3 2 0.55 20 0.3
- 2 0.2
- (a) Use a table of random numbers, or the random
number button on a calculator, to simulate the
operation of a machine for 12 months and hence
estimate a probability distribution for the
profit that would be lost per month if the
machine was purchased. - (b) Explain why the model is likely to be a
simplification of the real problem - Excel RAND(), IF
9- Chapter 8
- Revising Judgments
- in the Light of
- New Information
10 Bayes theorem Stats book
11Bayes theorem
Prior probability
New information
Posterior probability
Looks more like Bayesian estimation to me.
12The components problem fig. 8.1
13Applying Bayes theorem to the components problem
figure 8.2
14 Bayes Procedure
- 1) Construct a tree with branches representing
all the possible events which can occur and write
the prior probabilities for these events on the
branches. - 2) Extend the tree by attaching to each branch
a new branch which represents the new information
which you have obtained. On each branch write the
conditional probability of obtaining this
information given the circumstance represented by
the preceding branch. - 3) Obtain the joint probabilities by multiplying
each prior probability by the conditional
probability which follows it on the tree. - 4) Sum the joint probabilities.
- 5) Divide the 'appropriate' joint probability by
the sum of the joint probabilities to obtain the
posterior probability.
15Vague priors and very reliable information fig.
8.5
16The effect of the reliability of information on
the modification of prior probabilities Fig. 8.6
17The retailers problem with prior probabilities
fig. 8.7a
18Applying Bayes theorem to the retailers problem
fig. 8.7b
19Applying Bayes
theorem to the
20Determining the EVPI fig. 8.8
21Calculating the EVPI table 8.1
22Deciding whether to buy imperfect information
fig. 8.9
23If test indicates virus is present fig. 8.10a
24If test indicates virus is absent fig. 8.10b
25Determining the EVII fig. 8.11
26- Expected profit with imperfect
- information 62 155
- Expected profit without the
- information 57 000
- Expected value of imperfect
- information (EVII) 5 155
27EVII procedure p 233
- (1) Determine the course of action which would be
chosen using only the prior probabilities and
record the expected payoff of this course of
action - (2) Identify the possible indications which the
new information can give - (3) For each indication
- (a) Determine the probability that this
indication will occur - (b) Use Bayes' theorem to revise the
probabilities in the light of this indication - (c) Determine the best course of action in the
light of this indication (i.e. using the
posterior probabilities) and the expected payoff
of this course of action
28EVII procedure p 233
- (4) Multiply the probability of each indication
occurring by the expected payoff of the course of
action which should be taken if that indication
occurs and sum the resulting products. This will
give the expected payoff with imperfect
information - (5) The expected value of the imperfect
information is equal to the expected payoff with
imperfect information (derived in stage 4) less
the expected payoff of the course of action which
would be selected using the prior probabilities
(which was derived in stage 1).
29Ch 8 problem 4
- (4) A mining company is carrying out a survey in
a region of Western Australia. On the basis of
preliminary results, the company's senior
geologist estimates that there is a 60
probability that a particular mineral will be
found in quantities that would justify commercial
investment in the region. Further research is
then carried out and this suggests that
commercially viable quantities of the mineral
will be found. It is estimated that this research
has a 75 chance of giving a correct indication.
Revise the senior geologist's prior probability
in the light of the research results.
30(No Transcript)
31Ch 8 problem 6
32Ch 8 problem 6
33Ch 8 problem 6
34(No Transcript)
35(No Transcript)
36(No Transcript)
37(No Transcript)
38- Chapter 9
- Biases in
- Probability Assessment
39Chapter Outline
- I Introduction
- 15 questions
- II Hueristics Biases
- III The availability heuristic
- A Biases associated with availabilty heuristic
- 1 When ease of recall is not a associated with
probability - 2 Ease of imagination is not realted to
probability - Q1-Q3
- 3 Illusory correlation
40Chapter Outline
- IV The representativeness heuristic
- A Biases associated with representativeness
heuristic - 1 Ignoring base-rate frequencies Q4
- 2 Expecting sequences of events to appear
random Q5 - 3 Expecting chance to be self-correcting Q6-Q7
- 4 Ignoring regression to the mean Q8
- 5 The conjunction fallacy Q9-Q10
- V The anchoring and adjustment heuristic
- A Biases associated with anchoring and
adjustment - 1 Insufficient adjustment Q11
- 2 Overestimating the probability of conjunctive
events Q12 - 3 Underestimating probabilities for disjunctive
events Q13 - 3 (sic) Overconfidence Q14
41Chapter Outline
- VI Other judgmental biases
- 1 Believing desirable outcomes are more probable
- 2 Biased assessment of covariation Q15
- VII Is human probability judgment really so poor?
- 1 Subjects in studies may be unrepresentative of
real decision makers - 2 Laboratory tasks may be untypical of
real-world problems - 3 Tasks may be misunderstood by subjects
- 4 Subjects may be poorly motivated
- 5 Citation bias
- 6 Real-world studies suggest better performance
- 7 People think in frequencies not probabilities
42Heuristics and biases
- People use simple mental strategies or heuristics
to cope with difficult judgments - These heuristics are often efficient ways of
coping with complex problems, but... - They can also lead to systematically biased
judgments
43Three common heuristics
- The availability heuristic Vivid recall more
likely and vice versa - The representativeness heuristic Is Peter a
salesman? - Anchoring and adjustmentBase on know level
44The availability heuristic
- People assess the probability of events
- by how easily these events can be
- brought to mind.
- e.g. how easily they can be recalled or
- imagined
45Biases resulting from the availability heuristic
- Ease of recall may not be associated with
probability (Older crime victims) - Easily imagined events are not necessarily more
probable Q1-Q3 (Delays of the project like
strikes) - Illusory correlation (foreign supplier quality)
46The representativeness heuristic
- Used where people have to judge the probability
that - an object or person belongs to a particular
category - an event originates from a particular process
- The assessed probability is based on the extent
to which the object, person or event appears to
be representative of the category or process.
47Biases resulting from the representativeness
heuristic
- Ignoring base rate frequencies Q4 Engineer,
salesman judgment - Expecting short sequences of events to look
random Q5 Production miss plan - Expecting chance to be self-correcting Q6-Q7 hot,
cold lottery numbers - Ignoring regression to the mean Q8 SI cover
- The conjunction fallacy Q9-Q10 Linda
48Anchoring and adjustment
- People make estimates by starting from an
initial value (the anchor) and adjusting from it
to get their final estimate
49Biases resulting from anchoring and adjustment
- Insufficient adjustment from the anchor Q11 UN
membership - Overestimating the probability of conjunctive
events Q12 not multiplying - Underestimating the probability of disjunctive
events Q13 similar to above - Overconfidence Q14 too narrow CIs
50Other judgmental biases
- Believing desirable outcomes are more probable
My life will be better for housing and heart
attacks. - Biased assessment of covariation Q15
- table p 263
51Is human judgment really so poor?
- Subjects taking part in studies may not be
representative of management decision makers
undergrads - Laboratory tasks may be untypical of real-world
problems minor context changes important - These tasks may often be misunderstood by
subjects my correlation estimates (continu
ed)
52Is human judgment really so poor?
- Subjects in studies may be poorly
- motivated (not real)
- 5. Studies showing good performance tend to be
cited less often (is warning better?) - 6. Real-world studies suggest better
performance Experts are good NWS - 7. People think in terms of frequencies not
probabilities We are asking unnaturally, Linda
modified
53Choosing how to develop a subjective probability
assessment figure 9.12
54