Introduction to pyAgrum

pyAgrum is a scientific C++ and Python library dedicated to Bayesian networks (BN) and other Probabilistic Graphical Models. Based on the C++ aGrUM library, it provides a high-level interface to the C++ part of aGrUM allowing to create, manage and perform efficient computations with Bayesian networks and others probabilistic graphical models : Markov random fields (MRF), influence diagrams (ID) and LIMIDs, credal networks (CN), dynamic BN (dBN), probabilistic relational models (PRM).

Important

Since pyAgrum 2.0.0, the package name follows PEP8 rules and is now pyagrum (lowercase). Please use import pyagrum instead of import pyAgrum in your code.

See the CHANGELOG for more details.

The module is generated using the SWIG interface generator. Custom-written code was added to make the interface more user friendly.

pyAgrum aims to allow to easily use (as well as to prototype new algorithms on) Bayesian network and other graphical models.

pyAgrum contains :

• tutorials as jupyter notebooks.
• a comprehensive API documentation.
• a gitlab repository.
• and a website.

1- Fundamental components

• Graphs manipulation
  • Edges and Arcs
  • Directed Graphs
  • Undirected Graphs
  • Mixed Graph
  • Partially Directed Graph (DAG)
• Random Variables
  • Common API for Random Discrete Variables
  • Concrete classes for Random Discrete Variables
• Tensor and Instantiation
  • Instantiation
  • Tensor

2- Bayesian networks

• Model for Bayesian Network
  • BayesNet
• Input/Output Formats for Bayesian Networks
  • Supported Formats
  • Other Supported Formats
• Inference
  • Exact Inference
  • Approximated Inference
• Learning
  • Class BNLearner
• Classifier with Bayesian Networks
  • Classifier using Bayesian networks
• Tools for Bayesian networks
  • Random generation of Bayesian network
  • Generation of database
  • Information Theory of Bayesian network
  • Meek Rules Algorithm
  • Comparison of Bayesian networks
  • Explanation and analysis
  • Fragment of Bayesian networks

3- Causality

• Probabilistic causality
  • Causal models and do-calculus
  • Causel Effect estimation
• Tools for causality
  • Notebook’s tools for causality
  • Exceptions
  • Abstract Syntax Tree for Do-Calculus
• Causal Model
  • CausalModel
• Causal Inference (a.k.a. do-calculus)
  • causalImpact()
  • doCalculusWithObservation()
  • identifyingIntervention()
• Other functions for causal inference
  • CausalFormula
  • backdoor_generator()
  • frontdoor_generator()
• Causal Effect Estimation
  • CausalEffectEstimation

4- Other graphical models

• Influence Diagram and LIMIDS
  • Model for Decision in PGM
  • Inference for Influence Diagram
• Credal Network
  • CN Model
  • CN Inference
• Markov random field
  • Undirected Graphical Model
  • Inference in Markov random fields
• Probabilistic Relational Models
  • PRMexplorer

5- pyAgrum’s (experimentals) models

• Continuous-Time Bayesian Networks
  • The CTBN model
  • Inference in CTNB
  • Graphical tools for CTNB
  • Learning a CTBN
• Continuous Linear Gaussian Bayesian Networks
  • Gaussian Variables
  • The CLG model
  • Inference : Variable Elimination
  • Graphical tools for CLG
  • Learning a CLG
• Bayesian Network Mixture
  • Mixture Model
  • Inference on mixtures
  • Learning mixtures

6- pyagrum.lib modules

• Jupyter notebook with Graphical Models
• DiscreteTypeProcessor for graphical Models
• Export and visualize Graphical models and Tensors (png, pdf)
• pyagrum.lib.explain
• pyagrum.lib.dynamicBN
• other pyagrum.lib modules

7- pyAgrum’s tools

• Random numbers from aGrUM
• fast syntax
• Functions from pyagrum
• Listeners
• Exceptions from aGrUM

8- Customizing pyAgrum

• Configuration for pyAgrum

9- Appendices

• Bibliography
• Glossary