Insurance Bayesian Network Sample 4: A Discrete Medium-Scale Probabilistic Model

Insurance_4 is a discrete Bayesian network sample from the bnlearn repository, designed for modeling insurance-related variables. The network contains 27 nodes, 52 arcs, and 1008 parameters, with an average Markov blanket size of 5.19. It was authored by J. Binder, D. Koller, S. Russell, and K. Kanazawa, with a foundational citation from their 1997 Machine Learning paper on adaptive probabilistic networks.

Use Cases

• Benchmarking structure learning algorithms based on the known network topology of 27 nodes and 52 arcs.
• Testing probabilistic inference algorithms based on the network's 1008 parameters and discrete variable states.
• Studying the properties of medium-scale Bayesian networks based on metrics like average degree (3.85) and maximum in-degree (3).
• Educational demonstrations of Bayesian network concepts based on the well-documented Insurance repository example.

Strengths

• Network structure is explicitly defined with 27 nodes and 52 arcs, providing a clear topological benchmark.
• Detailed network statistics are provided, including 1008 parameters and an average Markov blanket size of 5.19.
• Has a clear academic provenance, being cited in a peer-reviewed 1997 Machine Learning paper by Binder et al.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for certain empirical tasks.
• Last update date is unknown; freshness unverified.

Provenance

Source

bnlearn Bayesian Network Repository (discrete-medium section), authors J. Binder, D. Koller, S. Russell, K. Kanazawa.

Collection Method

Likely a synthetic or engineered network for research and benchmarking purposes.

Time Range

null

Freshness

null

Geography

null