First-Order Theorem Proving: Subsampled Dataset with 10 Classes and 100 Features

A 2000-row, 100-column subsample of the first-order-theorem-proving dataset, generated with a random seed of 4. The subset contains up to 10 target classes and was created using a stratified sampling method. Author Eddie Bergman released it under a US public domain license on the OpenML platform.

Use Cases

• Benchmarking classification algorithms based on theorem-proving features.
• Analyzing feature importance for predicting theorem outcomes.
• Training models for automated reasoning tasks based on synthetic logical data.

Strengths

• Subsampling parameters are explicitly documented: seed=4, nrows=2000, ncols=100, nclasses=10.
• Stratified sampling was used, which likely preserves class distribution in the subset.

Limitations

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

Provenance

Source

OpenML (dataset ID 1475: first-order-theorem-proving)

Collection Method

Programmatic subsampling from a parent dataset using random selection and stratified train-test split.