A Bayesian nonparametric model for directed acyclic graphs

This code implements a Metropolis MCMC sampler for the Bayesian nonparametric prior on directed acyclic graphs (DAGs) by Patrick Dallaire et al., 2018. The npdag_simple_datasets.py script applies the method in conjunction with a simple and nearly parameter-free likelihood for binary classification.