Simulated Posterior Inclusion Probabilities for Modified BART and BKMR Models

Kaizong Ye published this dataset on figshare in May 2026. It contains simulated average Posterior Inclusion Probabilities (PIPs) for 15 exposures and a continuous outcome. The data is stored in a 9.5 KB XLS file.

Use Cases

• Compare variable selection performance between component-wise and hierarchical methods based on the described simulation.
• Benchmark modified Bayesian Additive Regression Trees (BART) against Bayesian Kernel Machine Regression (BKMR) based on the described metrics.
• Analyze the behavior of Posterior Inclusion Probabilities in a non-linear main effects only model based on the dataset's description.
• Validate new variable selection algorithms against established results from modified BART and BKMR.

Strengths

• The dataset is focused on a specific methodological comparison between modified BART and BKMR.
• It simulates a scenario with 15 exposures, providing a basis for multi-exposure analysis.
• The dataset is licensed under CC-BY-4.0, allowing for open sharing and reuse.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

figshare

Collection Method

Likely generated via statistical simulation, as described.

Freshness

Last updated 2026-05-11 17:51:46; freshness should be verified.