## Bayesian Network Repository

Several reference Bayesian networks are commonly used in literature as benchmarks. They are available in different formats from several sources, the most famous one being the Bayesian network repository hosted at the Hebrew University of Jerusalem. Others are shipped as examples of various Bayesian network-related software like Hugin or described in reference books such as Korb & Nicholson's “Bayesian Artificial Intelligence” or Koller & Friedman's “Probabilistic Graphical Models”.

Even more networks are available from various papers that used Bayesian networks to analyze data from various domains. This is especially true for Gaussian networks and conditional linear Gaussian networks, since the original Bayesian network repository included only discrete Bayesian networks.

Here I collected all the networks that I produced in my work, that I used for various simulations and
that I used to testing the implementations of `read.bif()`

, `read.dsc()`

and
`read.net()`

in **bnlearn**. All discrete networks are available in the BIF, DSC
and NET formats and have been quality-checked and fixed as needed (*i.e.* all conditional
probability distributions sum to one, no dummy nodes with a single level, no dangling dependencies on
non-existent nodes, *etc.*). R objects with the `bn.fit`

objects for all networks are
provided both as RDA and RDS files. RDA files can be loaded with *e.g.* `load("asia.rda")`

,
which creates an object called `bn`

in the current scope. RDS files can be loaded with
*e.g.* `asia = readRDS("asia.rda")`

, which returns the network and assigns it.

### Discrete Bayesian Networks

Small Networks (<20 nodes) | |||
---|---|---|---|

Name | Nodes | Arcs | Parameters |

ASIA | 8 | 8 | 18 |

CANCER | 5 | 4 | 10 |

EARTHQUAKE | 5 | 4 | 10 |

SACHS | 11 | 17 | 178 |

SURVEY | 6 | 6 | 21 |

Medium Networks (20–50 nodes) | |||

Name | Nodes | Arcs | Parameters |

ALARM | 37 | 46 | 509 |

BARLEY | 48 | 84 | 114005 |

CHILD | 20 | 25 | 230 |

INSURANCE | 27 | 52 | 984 |

MILDEW | 35 | 46 | 540150 |

WATER | 32 | 66 | 10083 |

Large Networks (50–100 nodes) | |||

Name | Nodes | Arcs | Parameters |

HAILFINDER | 56 | 66 | 2656 |

HEPAR II | 70 | 123 | 1453 |

WIN95PTS | 76 | 112 | 574 |

Very Large Networks (100–1000 nodes) | |||

Name | Nodes | Arcs | Parameters |

ANDES | 223 | 338 | 1157 |

DIABETES | 413 | 602 | 429409 |

LINK | 724 | 1125 | 14211 |

MUNIN (4 subnetworks) | 186–1041 | 273–1388 | 15622–80352 |

PATHFINDER | 135 | 200 | 77155 |

PIGS | 441 | 592 | 5618 |

Massive Networks (>1000 nodes) | |||

Name | Nodes | Arcs | Parameters |

MUNIN (full network) | 1041 | 1397 | 80592 |

MUNIN (4 subnetworks) | 186–1041 | 273–1388 | 15622–80352 |

### Gaussian Bayesian Networks

Medium Networks (20–50 nodes) | |||
---|---|---|---|

Name | Nodes | Arcs | Parameters |

ECOLI70 | 46 | 70 | 162 |

MAGIC-NIAB | 44 | 66 | 154 |

Large Networks (50–100 nodes) | |||

Name | Nodes | Arcs | Parameters |

MAGIC-IRRI | 64 | 102 | 230 |

Very Large Networks (101–1000 nodes) | |||

Name | Nodes | Arcs | Parameters |

ARTH150 | 107 | 150 | 364 |

### Conditional Linear Gaussian Bayesian Networks

Medium Networks (20–50 nodes) | |||
---|---|---|---|

Name | Nodes | Arcs | Parameters |

MEHRA | 24 | 71 | 324423 |