Matteo Bortoletto
Clelia Corridori
Michele Puppin
A Bayesian belief-network structure is a directed acyclic graph in which nodes represent domain variables and arcs between nodes represent probabilistic dependencies. Such network can be used for descriptive, diagnostic, predictive and prescriptive tasks. An efficient method for determining the relative probabilities of different belief-network structures, given a database of cases and a set of explicit assumptions, is the K2 algorithm. It can be used to learn the topology of a Bayes network, i.e. of finding the most probable belief-network structure, given a database. In this work we implement this algorithm in R language for a specific dataset. Furthermore we will use the bnstruct R package on the same dataset.