<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=648880075207035&amp;ev=PageView&amp;noscript=1">
 

Quick-Start Tutorial

"Bayesian networks are to probability calculus what spreadsheets are to arithmetic"

You can draw upon thousands of pages of documentation to learn about the details of the BayesiaLab software platform (see the BayesiaLab Library), and it may take you years to master it all. However, getting started with BayesiaLab and seeing meaningful results takes little time at all. All you need for this tutorial is a trial version of BayesiaLab (Win/Mac/Unix), which you can download here: www.bayesia.us/download.

Evidential Reasoning with Bayesian Networks

This Quick-Start Guide covers a simple example how you can model existing knowledge and then, given new information, compute inference by utilizing Bayes’ rule. The application of the famous Bayes’ rule itself is straightforward and wouldn’t necessarily require the use of Bayesian networks. However, the probability calculus required for applying Bayes’ rule would quickly become overwhelming for non-trivial problems. Bayesian networks, plus BayesiaLab’s inference algorithms, can elegantly handle all necessary computations. Also, given the counterintuitive results this example produces, it highlights the practical relevance of Bayesian networks as a support tool for probabilistic reasoning.

Fictional Case Study: Testing for a Rare Disease

A rare but fatal disease affects one in 10,000 adults. However, an inexpensive test exists that can detect this disease with great accuracy. More specifically, this test has a sensitivity of 99% (true positive rate) and a specificity of 99% (true negative rate). As part of a routine healthcare checkup, you are being tested for this disease. The test results are available instantly, but, unfortunately, your test result is positive, i.e., the test suggests that you have the disease. How concerned should you be at this point? Many would presumably think that there is a 99% probability that they do indeed have the disease. However, that is not a correct interpretation. In fact, this fallacy is so common that it is known as the "prosecutor's fallacy" or "the fallacy of the transposed conditional."

The following demo shows how to build a simple Bayesian network that can quickly provide the correct answer, thus avoiding the all-too-common fallacy.