Learning the Topology of a Bayesian Network Using K2 Algorithm
Prerequisites
Python versions supported:
Authors:
- Alberto Chimenti (University of Padova)
- Paolo Frazzetto (University of Padova)
- Vincenzo Schimmenti (University of Padova)
Brief Introduction
A Byesian belief-network structure
Using a Bayesian approach such a network can be constructed starting from a database which presents several records of the values combination of the system variables. One can find the most probable belief-network structure, given the dataset.
The K2 Algorithm
gives a computationally optimized way to search for the most probable structure.
Instead of maximizing the probability over all structures, one assumes that a node has no parents and incrementally adds that parent whose addition most increases the probability of the resulting structure.
A complete explanation of the methods and results obtained can be found in the notebook Project3_CFS_final.ipynb