User Guide#
Use these guides when you want workflow-oriented documentation before diving into the API reference. Each page focuses on a concrete task and links to the relevant examples and public APIs.
Core Workflow#
Learn causal graphs from data with constraint-based, score-based, and expert-guided algorithms.
Learn CPDs from data using MLE, Bayesian priors, or EM for missing data.
Compute posteriors, marginals, and MAP assignments with exact or approximate methods.
Sample observational, interventional, and conditional data from fitted models.
Causal Inference#
Determine identifiability using backdoor adjustment and frontdoor criteria.
Estimate treatment effects and interventional distributions from causal graphs and data.
Evaluation & Data#
Evaluate learned graphs with supervised and unsupervised metrics.
Curated benchmark datasets with ground-truth graphs and expert knowledge.
Ready-made networks from bnlearn, bnrep, and dagitty for benchmarking and exploration.
Utilities#
Save and load models in BIF, NET, XMLBIF, XDSL, and other formats.
Define graphs and CPDs directly for discrete, continuous, or dynamic models.
Visualize model structure with Graphviz, daft, and networkx backends.
Contributing#
Add new algorithms, datasets, and metrics using the built-in extension templates.