Pebl

Supports learning from both observational and interventional data, enabling causal inference as mentioned in the README's key features.

Handles incomplete datasets and latent variables using exact and heuristic methods, which is crucial for real-world data challenges.

Provides built-in learning algorithms and makes creating new ones simple, allowing for customization and research advancements.

Offers transparent parallel execution across clusters and cloud resources, facilitating large-scale computations as highlighted in the README.

Calculates edge marginals and consensus networks to assess model reliability, aiding in robust analysis.