Insurance 10: A Bayesian Network Sample for Insurance Risk Modeling

A sample from the Insurance Bayesian Network, a probabilistic model with 27 nodes and 52 arcs. The dataset contains 1008 parameters and was created by J. Binder, D. Koller, S. Russell, and K. Kanazawa, referenced in their 1997 Machine Learning publication.

Use Cases

• Learning and benchmarking Bayesian network structure learning algorithms based on the provided network topology.
• Testing probabilistic inference algorithms using the model's 1008 parameters.
• Studying the application of hidden variable models in insurance domains, as suggested by the source paper's title.
• Demonstrating risk prediction models in an insurance context.

Strengths

• Well-defined network structure with 27 nodes and 52 arcs, providing a concrete model for analysis.
• Substantial number of parameters (1008) for a discrete medium-sized network.
• Clear academic provenance, authored by notable researchers and cited in a 1997 Machine Learning journal paper.

Limitations

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

Provenance

Source

bnlearn Bayesian Network Repository, discrete-medium category.

Collection Method

Likely a synthetic or processed sample derived from the full Insurance Bayesian Network model.

Time Range

null

Freshness

unknown

Geography

null