The Murder Network

Startpage
Tutorial
Hello Bayes

The following explains the probability tables used for the two chance nodes Murderer? and DNA-match? used in the murder network.

The Murderer? node

In the presented case, Colombo knew for sure that the killer was member of a well defined group. In our model we assume that the killer is known to be from Denmark. The Danish population is approximately 5 millions. By picking a random individual from the street the probability of him (or her) being the murderer becomes as in Table 1.

Murderer?="yes" Murderer?="no"
1 4,999,999
Table 1: Cpt of the Murderer? node.

The DNA-match? Node

DNA typing of biological material has become one of the most powerful tools for personal identification in criminal investigations. The problem with DNA analysis lies in determining whether two DNA-fingerprints match. A match conclusion is therefore always accompanied by a probability of obtaining a match as good as that found in the evidence. This probability depends highly on the quality of the biological stain recovered from the scene of the crime but also on the frequency of the elements in the fingerprint occuring in the population. This indicates that even if a person is innocent his genetic fingerprint can be "similar" to the one extracted from the biological stain found at the scene of the crime.

In our model we assume that if the individual really is the killer the DNA fingerprint will match. Colombo is informed by the forensic experts that the quality of the recovered stain results in a 1/100000 chance of obtaining a coinciding match. Hence the probabilities for the node are given as in Table 2 .

  Murderer?="yes" Murderer?="no"
DNA-match?="yes" 1 1
DNA-match?="no" 0 99999
Table 2: Cpt of the DNA-match? node.

HUGIN Expert A/S, 2007 - comments to activex@hugin.com