bnlearn’s documentation!

bnlearn is for learning the graphical structure of Bayesian networks in Python! What benefits does bnlearn offer over other bayesian analysis implementations?

• Build on top of the pgmpy library

• Contains the most-wanted bayesian pipelines

• Simple and intuitive

• Focus on structure learning, parameter learning and inference.

Step-by-step guide for structure learning Step-by-step guide for knowledge-driven models

Content

Quickstart

• Quickstart
  • Installation (short)
  • Quick Examples

Introduction

• Introduction
• Installation
• Validate
• Uninstalling

Structure learning

• Structure learning
• Exhaustivesearch
• Hillclimbsearch
• Constraint-based
• Chow-liu
• Tree-augmented Naive Bayes (TAN)
• NaiveBayes
• Examples Structure learning
  • Example (1)
  • Example (2)

Parameter learning

• Parameter learning
• Maximum Likelihood Estimation
• Bayesian Parameter Estimation
• Examples Parameter learning
  • Example (1)
  • Example (2)

Inference

• Inference
• Inference Algorithms
• Examples Inference
  • Example (1)
  • Example (2)
  • Example (3)

Predict

• Predict

Other functionalities

• Directed Acyclic Graphs
  • Overview
  • Building a causal DAG
  • The causal DAG as a generative representation of joint probability
  • Specifying the probability distributions on your own
  • Inference on the causal generative model
• Sampling and datasets
  • Datasets
  • Example Sampling
• Black and white lists
• Topological sort
• Saving and Loading

Examples

• Examples
  • Example with DataFrames
  • Example with DAG
  • Import from BIF
  • Start with RAW data
  • Adjacency matrix
  • Create a Bayesian Network, learn its parameters from data and perform the inference
• Use Case Medical domain
  • Description
  • Import data
  • Make inferences when you have data and know-how
  • Determine causalities when you have data
  • Make inference when you have data

Plot

• Interactive plot
• Static plot
• Node properties
• Edge properties

API References

• bnlearn.structure_learning
• bnlearn.parameter_learning
• bnlearn.inference
• bnlearn.bnlearn

References

• Probabilistic Graphical Models using pgmpy

• Causality, Pearl, 2009, 2nd Editing

• If correlation doesn’t imply causation, then what does? from Michael Nielsen

• Lecture notes from Jonas Peters

• Elements of Causal Inference

• Causality slides

Related Packages

• Causal graphical models

• Causality

• Causal Inference

Source code and issue tracker

Available on Github, erdogant/bnlearn. Please report bugs, issues and feature extensions there.

Indices and tables

• Index

• Module Index

• Search Page