H2PC Algorithm

Overview

H2PC is a hybrid structure learning algorithm that combines a constraint-based phase with a score-based phase. In the first phase it uses the PC (Parents and Children) local discovery algorithm to learn the skeleton — identifying the direct neighbours of each node via conditional independence tests. In the second phase it uses hill-climbing with the BIC (or other chosen) score to orient the edges and refine the structure within the constraint skeleton.

By restricting the score-based search to the skeleton identified constraint-based phase, H2PC dramatically reduces the search space compared to running HC on the full graph. This makes it faster and less prone to over-fitting on high-dimensional datasets, while the score-based refinement typically produces better-oriented graphs than a purely constraint-based approach.

Class: hybrid · DAG Package: bnlearn

Reference

Gasse M., Aussem A. & Elghazel H. (2014) – A Hybrid Algorithm for BN Structure Learning with Application to Multi-Label Learning. Expert Syst. Appl. 41(15), 6755–6772. https://doi.org/10.1016/j.eswa.2014.04.032

Hyperparameters

Hyperparameter	Type	Default	Values	Description
alpha	float	0.05	—	p-value threshold below which a CI test indicates conditional independence.
ci_test	str	mi	mi, x2	Conditional independence test used in constraint-based learning.
iss	float	1.0	—	Imaginary Sample Size weighting the prior in Bayesian scores.
max_elapsed	int	No limit	—	Maximum allowed execution time in seconds.
max_iterations	int	No limit	—	Maximum number of iterations.
penalty_weight	float	1.0	—	Weight of the penalty component in AIC and BIC scores.
score	str	bic	aic, bdeu, bge, bic, k2, loglik	Scoring function for score-based learning.

Variants

Variant	Package
bnlearn	bnlearn