Sourced from pgmpy's releases.

Release v1.1.2

Please refer Changelog for details: https://github.com/pgmpy/pgmpy/blob/dev/CHANGELOG.md

Thanks to all the contributors.

@​AjayBora002 @​anusa-saha @​Anushka-2406 @​balgaly @​codejasleen @​Cyberpunk-San @​hanara2112 @​kajal-jotwani @​Nimish-4 @​officialasishkumar @​PewterZz @​SayedGameaSayed @​Vidhu-sri

Sourced from pgmpy's changelog.

[1.1.2] - 2026-04-30

Added

1. Prioritization option for PC orientation to control rule application order (#3363).
2. Effect sizes for CI tests (#3358).
3. Caching support to CI tests for faster repeated queries (#3349).
4. New CI tests (#3344).
5. Estimator argument to Pillai Trace CI test (#3072).
6. DAG.get_random can now generate graphs with a fixed number of edges (#3086).
7. Causal stats in DAG.get_stats() method (#3143).
8. Exposes estimators as attributes in residual based CI tests (#3367).

Changed

1. Optimizes power_divergence test and PC algorithm for better performance (#3356).
2. Refactors parameter estimation methods into a dedicated pgmpy.parameter_estimator package (#3325).
3. Refactors SHD to accept covariant arguments in __init__ for composability (#3310).
4. Improves deprecation warning messages (#3343).
5. Uses combinations instead of permutations in _orient_colliders for better performance (#3195).
6. Revamps documentation website (#2615).
7. Reject negative probability values in is_valid_cpd (#3052).

Fixed

1. Fixes bug in collider orientation for PC (#3362).
2. Fixes edge orientation in GES (#3304).
3. Fixes GES forward/backward/turning phases to be deterministic across hash seeds (#3303).
4. Fixes __eq__ to check structure equality in LinearGaussianBayesianNetwork (#3276).
5. Fixes incorrect has_missing_data tag for tubingen dataset (#3231).
6. Fixes minor typo in ValueError messages in state_name.py (#3309).
7. Fixes failing doctests in BIF, NET, XMLBIF, and UAI readwrite modules (#3226, #3230, #3262, #3263).

[1.1.1] - 2026-04-30

Version skipped due to errors in build and tagging.

[1.1.0] - 2026-04-01

Added

1. New role-aware causal graph infrastructure, including PDAG, ADMG, MAG, AncestralBase, and SimpleCausalModel.
2. New graph utilities in DAG, including to_pdag, to_dagitty, to_lavaan, public get_ancestors, edge_strength, get_stats, and __hash__.
3. DAG.to_daft and DAG.to_graphviz can now annotate plots with computed edge strengths.
4. DAG.from_dagitty can now construct LinearGaussianBayesianNetwork instances when the dagitty model includes beta coefficients.
5. New sklearn-compatible causal discovery package pgmpy.causal_discovery with refactored PC, GES, HillClimbSearch, and ExpertInLoop estimators.
6. ExpertKnowledge now supports search_space, richer string representations, temporal-order handling, and tighter integration with causal discovery and ExpertInLoop.
7. New causal effect APIs built around pgmpy.identification, including shared identification base classes and adjustment/frontdoor workflows.
8. New pgmpy.prediction module with sklearn-compatible causal prediction estimators: NaiveAdjustmentRegressor, DoubleMLRegressor, and NaiveIVRegressor.
9. New pgmpy.ci_tests package with class-based CI tests and registry-based lookup, including FisherZ, PearsonrEquivalence, and estimator-configurable GCM.
10. New pgmpy.structure_score package and new causal graph evaluation metrics: AdjacencyConfusionMatrix and OrientationConfusionMatrix.
11. LinearGaussianBayesianNetwork now supports JSON load/save, log_likelihood, predict_probability, improved predict, and richer simulation features for interventions, evidence, virtual interventions, and missing-data generation.
12. FunctionalBayesianNetwork sampling now supports vectorized FunctionalCPD functions and interventional simulation.
13. ExpectationMaximization gained apply_smoothing, configurable init_cpds, and support for node-specific priors.
14. DynamicBayesianNetwork.simulate gained a return_format argument.
15. New dataset and example-model registries, pgmpy.datasets and pgmpy.example_models, with filterable catalogs and on-demand loading.
16. Many new benchmark datasets were added, including Boston Housing, Wine Quality, Galton Stature, Seoul Bike, Pima Diabetes, Student Performance, South German Credit, Cystic Fibrosis, Yacht Hydrodynamics, HTRU2, Myocardial Infarction, IQ Brain Size, Contraceptive Method, Pittsburgh Bridges, Residential Building, Hitters, Superconductivity, Lead, Cities, College Plans, Depression Coping, US Crime, Hungary chickenpox, the Angrist-Krueger 1980 census extract, dropout covariance data, and the Tuebingen pairs.

... (truncated)

• 617cb48 Release v1.1.2
• 3c01db9 Skip test if sempler not in environment
• 34a750a Release v1.1.1 (#3368)
• 5cdf981 Exposes estimators as attributes in residual based CI tests (#3367)
• b1a30db Adds prioritzation option for PC orientation (#3363)
• 5484569 Fixes bug in collider orientaiton for PC (#3362)
• 4986030 Adds effect sizes for CI tests (#3358)
• 9caf211 Optimizes power_divergence test and PC algorithm (#3356)
• d368eb2 Refactor parameter estimation methods (#3325)
• 5239dfe Adds caching support to CI tests (#3349)
• Additional commits viewable in compare view

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore <dependency name> major version will close this group update PR and stop Dependabot creating any more for the specific dependency's major version (unless you unignore this specific dependency's major version or upgrade to it yourself)
• @dependabot ignore <dependency name> minor version will close this group update PR and stop Dependabot creating any more for the specific dependency's minor version (unless you unignore this specific dependency's minor version or upgrade to it yourself)
• @dependabot ignore <dependency name> will close this group update PR and stop Dependabot creating any more for the specific dependency (unless you unignore this specific dependency or upgrade to it yourself)
• @dependabot unignore <dependency name> will remove all of the ignore conditions of the specified dependency
• @dependabot unignore <dependency name> <ignore condition> will remove the ignore condition of the specified dependency and ignore conditions