Chapter 11Bayesian decision networks

One of the main threads of argument pursued so far in this book considers Bayesian networks as a widely and flexibly applicable formalism that supports sound reasoning when the available information is incomplete. On the basis of observations, of the kind typically encountered in forensic science, one can construct probabilistic arguments to propositions of interest. As such, Bayesian networks provide a coherent environment wherein beliefs about target propositions can be revised upon receipt of newly acquired information. This represents a fundamental preliminary requirement for an additional step, that is, the coherent use of beliefs in action. This is also more commonly known as decision making under uncertainty (Lindley 2006). Chapter 2 showed that Bayesian networks can be extended to incorporate the basic ingredients necessary to perform so-called Bayesian decision analyses. In essence, this extension consists of decision and utility nodes (Section 1.1.9). They represent, broadly speaking, actions available to the scientist—or, in a wider sense, to the decision maker—and values for possible consequences of these actions, respectively. This chapter emphasizes the relevance of these extensions for forensic science by focusing on selected examples.

11.1 Decision making in forensic science

There are many situations in forensic science in which decisions need to be made under circumstances of uncertainty. It is routinely asked, for example, ...

Get Bayesian Networks for Probabilistic Inference and Decision Analysis in Forensic Science, 2nd Edition now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.