Bayesian Network CPT calculator Written to assist in a research project experiementing with Bayesian network models. Given a sample dataset and a bayesian network learned from that dataset, this code aims to isolate datapoints that differ significantly from the others, using the connections found from the bayesian network. The code does not do any bayesian network learning. For the research project, I relied on R's wonderful bnlearn package to learn bayesian networks. I wrote a DSC parser that would build NetworkX objects from the networks that I exported from R. File Descriptions: cpt.py: Calculates conditional probability tables and attaches them to their respective nodes in the bayesian network. dsc.py: Import DSC files (a network exchange format). harm.py: A heuristical optimization algorithm (harmony search) to try an learn a simple bayesian network. This didn't work as well as I had hoped. Networks generated were not very good. Switched to using bnlearn package from R. learn.py: An attempt at a simple greedy network learner. network.py: Representation for a bayesian network. Bayesian networks are directed acyclic graphs. This object subclasses networkx.DiGraph. The class also has special methods related to my research for retrieving the cpt and calculating joint probabilities. reader.py: Read the output of R. I ended up using modelstrings from R to exchange the networks between R and Python. score.py: Scoring algorithm used in the greedy algorithm and harmony search util.py: various utility functions to do a collection of useful things.