AI summary by TwinMind
Unit 8: Decision Analysis 2, Video 2: Decision Trees
MIT OpenCourseWare
8m 0s
Decision Analysis General Features
- Provides structured, efficient way to help people recognize complexity and probabilities
- Offers simple way of defining choices
- Assumes everything is discrete choice (yes/no, yes/somewhat/a lot/maybe not at all)
- Forces thinking in terms of discrete choices, not continuum
- Enables/compels looking over several periods
- Not just first period decisions
- Example: MIT course selection - first semester choices enable/prevent future options due to prerequisites
- Standardized method with available tools
- Structured process like Excel
- Provides widespread language for decision thinking
- Can include non-numerical results beyond income/present value
- Handles risk aversions, consumer satisfaction, utility concepts
- Allows weighting against bad outcomes (e.g., retirement savings example: average $1M could be either $2M or $0 - $0 outcome creates retirement unhappiness)
Decision Tree Structure
- Central element of decision analysis
- Shows choices over several periods
- Four key elements: choices, outcomes, structure, data
- Two main components:
- Decision points (left side - starting choices)
- Chance outcomes (results after decisions)
- Basic sequence: decision → chance → decision → chance
- Tree structure metaphor: lying on side, root at start, branches extending outward
Decision Tree Complexity Example
- Simple 2-choice, 2-outcome, 2-stage process = 16 total outcomes
- Complexity explosion potential:
- 5 choices × 5 outcomes = 25 possibilities
- Repeated stages: 25² = 625 outcomes
- Chess game example:
- 10-12 possible opening moves (pawns, rooks, castles)
- Opponent responses multiply complexity
- Problems can "explode" quickly
Method Limitations
- Decision analysis good for simple problems
- Complicated problems become too messy to manage effectively
