|
BN /
SamplesSample Bayesian NetworksThese are networks used in the tutorial. Some of these are answers to exercises. All of the networks can be found at: http://pluto.coe.fsu.edu/BNinEA/Examples/SampleBayesianNetworks/ (tarball, zip) Dancer networkThis is a potential answer to a question which asks the participants to draw a Bayesian network for several dancers each of which give several performances, each of which are judged by several raters. ![]() Note: This example is a graphical structure only, no probabilities. Drivers licence networkAn activity we often do during the tutorial is to have the participants build a network for a driver's licence examination (or a subset of that examination). This was a solution from one of the participant groups. ![]() Note: This example is a graphical structure only, no probabilities. Accident PronenessThis example is taken from Feller (1968) Invalid BibTex Entry! and shows that even though accidents happen independently in each year, they are not independent because of the hidden driving proficiency variable. This illustrates the idea of conditional independence?. ![]() HIV TestThis is an example of what is known as the rare disease problem. A common fallacy when evaluating evidence is to just look at the evidence and ignore the base rate. This comes up particularly in screening tests for rare diseases. The source of the numbers in this example are given in Chapter 3 of Bayesian Networks in Educational Assessment. HIV test prior to seeing evidence. ![]() HIV test after seeing positive evidence. ![]() HIV test after seeing negative evidence. ![]() Combination of two skillsThis little example explores the difference between three different design patterns for combining the evidence from two skills: Compensatory?, Conjunctive? and Disjunctive?. There are three independent networks, each with the same marginal probabilities but different conditional probability tables for the observation node. Details can be found in Chapter 6 of Bayesian networks in Educational Assessment. ![]() Some things to try with this network:
|