Update 11/15/2011

Pebl source code and issues are now hosted at http://github.com/abhik/pebl. This site is only for historical purposes.

Pebl is a python library and command line application for learning the structure of a Bayesian network given prior knowledge and observations. Pebl includes the following features:

• Can learn with observational and interventional data
• Handles missing values and hidden variables using exact and heuristic methods
• Provides several learning algorithms; makes creating new ones simple
• Has facilities for transparent parallel execution using several cluster/grid resources
• Calculates edge marginals and consensus networks
• Presents results in a variety of formats

Pebl Documentation includes installation instructions, tutorial and API reference.

Pebl has been developed at the Systems Biology Lab at the University of Michigan and is available with a permissive MIT-style license.

Update 3/6/2009

Pebl is now published in the Journal of Machine Learning Research. Please cite the paper if you use Pebl for your work. Abstract and PDF