MS&E 107 Lecture Notes - Lecture 11: Queueing Theory, Decision Tree Learning, Sensitivity Analysis
Document Summary
Same servicing and arriving rate - queue length is zero! Uncertainty branch: displays sum product of each probability times each outcome weight. Flaw of average potential - the sum product is the average (probability * weight) Ex: decision - stick in the eye or good time, with chance of getting caught. Sensitivity analysis: play with probability of getting caught. 10% or higher (at 20% of getting caught) - worth on average -. Decision: still go for the good time because -40 is better than losing -300. 50% chance of getting caught - stick in the eye. Try not to estimate critical probability (getting caught) - model building. Value of perfect information: pay to get correct information. What is is worth to find out the perfect information? (ex: night watchman randomly guards a building, brother in law knows watchman) Decide - find out situation changed to find out - decide.