Flashcards
Decision Theory
Cards in this deck (26)
- 1Aleatoric UncertaintyStatistical uncertainy, from natural randomness e.g Earthquakes, Human Behaivour etc.
- 2Epistemic UncertaintySystematic uncertainty that comes from inaccuracies in understandings and models.
- 3The three options of action with uncertainties- Ignore
- Allow for using intuition (ie Engineering Judgement)
- Scientific approach by using probabilistic and statistical analysis - 4Null HypothesisA general statement or default position that there is no relationship between two measured phenomena
- 5Type I ErrorNull Hypothesis is correct but is rejected - false positive
- 6Type II ErrorNull Hypothesis is false but is not rejected - false negative
- 7How to fix type I and type II errors respectivelyType I - Increase your level of confidence (more precise)
Type 2 - Descriptive testing (more detailed)
##clarify later## - 8Risk analysisA technique used to identify and assess factors that may jeopardize the success of a project or achieving a goal.
- 9Risk ManagementIdentifying the probability and consequence of making the wrong decisions
- 10State of Nature (SON)The true value of an uncertain variable.
It can never be determined with absolute confidence. - 11How to get and refine a SONYou can estimate the SON through prior knowledge, and obtain additional information to refine it by doing experiments
- 12Statistical Decision TheoryIdentifies the optimal decision when given a probability distribution for a state of nature
- 13Preposterior AnalysisWhen there are several possible tests identifying which one needs to be done. Tests should only be done id expected value of information is greater than obtaining cost.
- 14Stages of Posterior Analysis2 Stages in Decision Tree - Action and State
- 15Stages of Preposterior analysis4 Stages- Test (e), result (z), action(a), state (theta)
- 16Allais' ParadoxActual observed choices are inconsistent with predictions from expected utility theory
- 17Rational Decision-making approachMaking decisions purely based on probability and consequence of each possible outcome
- 18UtilityEssentially the benefits something gives
- 19Robust numerical measure of preference requirements- Has a scale ordering expected values
Reflects decision makers subjective preferences - 20Standard GambleAllows subjectivity, whilst preserving order of expected values.
It is based on a decision tree, one leading to a certain outcome, the other a gamble. - 21Three different types of decision making criteria (alternatives to rational approach)- Criteria of Pessimism
For each action find minimum utility. Choose action with largest
- Criteria of Optimism
For each action find maximum utility.
Choose action with largest
- Criterion of regret
Make table of regret, choose action with smallest maximum regrt - 22ReliabilityProbability that component (or system) will function properly
- 23Assumptions made in reliability engineering- All components are independent
- No partial failures
- Known reliabilities per component - 24Types of systemsSeries - System survives if all components survive
Parallel - System survives if not all components fail - 25Birnbaum ReliabilityImprove the link with highest Reliability Importance
In series - Improve the weaker link
In parallel - Improve the stronger link - 26Henly and KumamotoImprove link with highest Critical Importance
In series - Either or
In parallel - Improve weaker link