BU275 Lecture Notes - Lecture 2: Decision Tree Learning, Conditional Probability
Document Summary
Decision nodes: squares, branches out of these can be changed/chosen, not predetermined, not all-inclusive. Event nodes: circles, states of nature, outcomes, all inclusive, usually have a probability attached, uncontrollable. Leaves: diamond, at the end of a string of decisions, sums up the sequence of branches/decisions and outcomes. Find out if there is oil or not. Calculating the expected value of a decision/outcome path: Trace back the paths and sum the costs/payoffs that have occurred. Ev of the decision nodes is the maximum profit value out of the options, ev of the event nodes is the expected value of occurrence. If you didn"t buy the test: expected value of drilling = (. 25*700,000)+(. 75*-100,000) = ,000 expected value of selling = ,000 total expected value of not buying the test is the maximum between. ,000 and ,000 so the expected value of not buying the test is.
