The HPBNET Procedure

The HPBNET procedure is a high-performance procedure that learns a Bayesian network from an input data set. A Bayesian network is a directed acyclic graphical model in which nodes represent random variables and the links between nodes represent conditional dependency of the random variables. Because the Bayesian network provides conditional independence structure and a conditional probability table at each node, the model has been used successfully as a predictive model in supervised data mining. For more information about Bayesian networks, see Pearl (1988).

The HPBNET procedure can learn different types of Bayesian network structures, including naive, tree-augmented naive (TAN), Bayesian network-augmented naive (BAN), parent-child Bayesian network, and Markov blanket. PROC HPBNET performs efficient variable selection through independence tests, and it selects the best model automatically from the specified parameters. It also generates SAS DATA step code to score a data set.