Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Tecnico, Decision support models > T2, Decision trees, influence diagrams & payoff > Flashcards

T2, Decision trees, influence diagrams & payoff Flashcards

1
Q

Which are the key questions regarding uncertainty?

A
  1. what are the key uncertainties?
  2. What are the possible outcomes of these uncertainties?
  3. What are the chances of occurrence of each possible outcome?
  4. What are the consequences of each outcome?
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is a risk profile?

A

A table containing all different:

  1. outcomes
  2. chances
  3. consequences
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Which is the most classical approach to model decision problems involving sequential decisions under uncertainty?§

A

Decision tables & Decision Trees

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Vad är idéen bakom decision trees?

A

Att konsekvenserna av ett beslut kan avgöras av ett antal externa faktorer som inger utom beslutsfattarens kontroll

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Vilka delar består ett decision tree av?

A
  1. Decision nodes
    - -> representerar beslut
    - -> en förgrening från en decision node måste vara på sådan form att beslutsfattaren endast kan ta ett av besluten
    - -> fyrkanter
  2. Chance nodes
    - -> representerar slumpen, osäkra händelser
    - -> måste ha förgreningar som motsvaras av ett set likvärdiga konsekvensalternativ
    - -> ringar
  3. Konsekvens nod
    - -> är alltid i änden på trädet
    - -> när osäkerheten är löst kan endast en av konsekvenserna uppstå
    - -> trekant
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

vad är ett decision tree?

A

ett beslutsträd representerar alla möjliga vägar som beslutsfattaren kan gå och inkluderar alla möjliga beslutsalternativ och konsekvenser av chanser

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

vad innebär det om man har en beslutsnod innan en chance nod ?

A

Att beslutet måste tas i väntan på chance-händelsen

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

vad innebär det om man har en chancenod innan en beslutsnod ?

A

Beslutet är villkorat av utfallet av chancenoden.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

what is an influence diagram?

A

A diagram where nodes ate put together in a graph connected by arcs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

what does arcs represent in a influence tree?

A

relationship

relevance or sequence

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

vad menas med the value of information?

A

beskriver värdet av att ta reda på utfallet av chansnoden innan du måste fatta ett beslut.

  • Informationen har inget värde om den inte påverkar dina handlingar
  • värdet bestäms av den förbättringen som den bidrar med jämfört med utan förbättringen
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

vad innebär höga values of information?

A

representerar de bästa möjligheterna att förbättra det förväntade värdet

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

vad innebär låga chances of information?

A

chanshändelser med låga values of information är förmodligen inte värt att gå vidare med

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

vad innebär value of control?

A

Value of control för en händelse beskriver värdet av att ha möjligheten att bestämma utfall av “the uncertainty” istället för att chansa (gå till chansnod)

  • Värdet kommer ifrån att kan kan garantera det mest önskvärda utfallet och förebygga de minst önskvärda utfallen.
  • Most favorable outcome- least favorable outcome
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

When is it good to use value of control?

A

for example when you have a variable with high Information value but that is hard to predict

–> for example weather

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the the greatest opportunity for improving your outcomes?

A

To gain high value of control over chance events

–> expencive

17
Q

nämn några vanliga källor till value of control!

A
  1. Mer personal, tid, pengar eller andra resurser
  2. Pr eller reklam
  3. försäkring
18
Q

nämn några vanliga typer av imperfekt kontroll:

A
  1. kontroll förbättrar bara sannolikheten

2. man kan inte välja det bästa

19
Q

Nämn några viktiga lärdomar om value of control!

A
  1. Value of control kan komma från att kontrollera underliggande osäkerhet
  2. Value of control kan komma från att isolera sig själv från effekterna av osäkerheten
  3. value of control är vanligen större än eller lika med värdet av informationen
20
Q

Vilka styrkor och svagheter finns det med influence diagram?

A

STYRKOR

  • kompakt
  • bra för kommunikation i struktureringsfasen
  • bra för att överblicka större problem
  • bra för att förstå relationen mellan osäkerhetsnoder

SVAGHETER
- inte så detaljerat verktyg

21
Q

Vilka styrkor och svagheter finns det med decision treees?

A

STYRKOR

  • visar detaljer
  • bra för djupare förståelse
  • flexibel representation
  • bäst för asymmetriska problem i beslutsfattande
  • lämplig modell för känslighetsanalys

SVAGHETERA
- Blir stökigt för stora problem

22
Q

What does it mean with imperfect information?

A

Decision maker waits for information before making a decision

23
Q

What does EVII stand for?

A
  • Expected value of imperfect information
  • EVII=EMV waiting for survey results - EMV for deciding without surveys
  • You don’t want to pay more than the EVII for the survey
24
Q

How do you calculate EMV?

A

ProbabilityChoice1_outcome1+probabilitychoice1_outcome2

25
Q

Why is a sensitivity analysis good?

A
  1. You can see the indifference point and the probability for that point
    2.
26
Q

What is EVPI?

A
  • Expected value of perfect information
  • the maximum amount that the DM should be willing to pay for perfect information
  • EVPI=EMV after survey results - EMV for deciding without surveys (????)
27
Q

Vad handlar graham’s decision problem om?

A

Att man viktar egenskaper i ditt beslutsträd

28
Q

what is a event tree? applications?

A

A decision res without any decisions

APPLICATION
–> risk analysis of technological systems
identifications of improvements in protection systems

29
Q

what is an accidental event?

A

the first significant deviation from a normal situation that may lead to unwanted consequences

30
Q

what is a fault tree?

A

A fault tree begins with an initial system problem and then represents all the corrective actions or system events that can be taken to correct the default

31
Q

applications for a defaut tree?

A
  1. Reliability and safety analysis
    - -> failure of an item a system is often caused by failure of other items
    - -> method of breaking down these chains of failures
    - -> identifying combinations of faults that causes other faults
    - ->
32
Q

What are main types of the combinations of faults?

A
  1. where sevaral items must fail together to cause another item to fail (AND-gate)
  2. where only one of a number of possible faults need to happen to cause another item to fail. (OR-gate)

These combinations work as gates by preventing the failure event to happen if specific conditions are met

33
Q

What are the steps in fault tree analysis in problem solving?

A
  1. identify
  2. define
  3. problem
  4. cause
  5. solution
  6. implement
  7. review
  8. follow up
34
Q

vad är fault trees useful for?

A
  1. When the effect of a failure is known.
  2. To find how a certain effect might be caused by combination of other failures
  3. When designing a solution, to identify ways it may fail and consequently find ways of making the solution more robust
  4. To identify risks in a system and consequently identify risk reduction measures
  5. use to find failures which can cause the failure of all parts
35
Q

When is a payoff table useful?

A
  1. In situations when there is uncertainty but no probabilistic information
  2. When there are several strategies and several scenarios
36
Q

What are the different algorithms to make choices under uncertainty?

A
  1. Laplace
    - -> principle of insufficient reason
  2. Wald
    - -> Pessimist - Maximin
  3. MaxiMax
    - -> optimist
  4. Hurwicks
    - -> intermediate
  5. Savage
    - -> MiniMax regret
37
Q

Name some complementary concepts for calculating expected monetary value and risk profile!

A
  1. The risk Profile concept
    - -> A risk profile is a graph that shows the chances associated with possible consequences
    - -> each risk profile is associated with a strategy, a particular immediate alternative and specific alternatives in future decisions
  2. The cumulative risk profile concept
    - -> The vertical axis is the chance that the payoff is less than or equal to the corresponding value on the horizontal axis
    - -> horisontal axis shows the chance that the payoff will be less than or equal to that specific value
    - -> results from adding up, or accumulating the chances of the individual payoffs
  3. The expected value concept
    - -> E(y)= sum of yi*pi from I=1 to n
    - -> random variable y
    - -> expected value: best quess for y
    - -> if you were able to observe many outcomes of calculated average of all the outcomes it would be close to E(y)

Tecnico, Decision support models (16 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions