Introduction

It is expected to build a Bayes net structure, fit it to data, and report on its performance. You can use either Netica software or the pyAgrum python package for implementation.

You can get Netica free from this link:

https://www.norsys.com/netica.html

Note that the free version of Netica is "full featured, but doesn't work with large models.For instance, it can't save a network having more than 15 nodes, and can't read one with more than 60 nodes.Also, during learning, it can only learn from 1000 or fewer cases at a time."

Information about the pyAgrum package is available here: https://agrum.gitlab.io/pages/pyagrum.html

Requirements

1. Find a suitable dataset online that would be good for modeling with a Bayes Net. Note that you may have to downsample this data due to limitations of the free Netica software.

2. Make a Bayesian Network structure that attempts to model the data. You will need to think about how to best form the structure. You may also try using algorithms to learn the structure. You should try at least two different structures.

3. For each Bayes Net structure you implement, learn the network parameters and get the error rate (In Netica, use "learn from cases"). You can try different learning algorithms.

4. Perform some example inferences.

5. Make a report describing your process and discuss the results.