Unified Analysis of ML Performance, Interpretability, and Statistical Tests

Dan Jiang's 5.5 KB Excel file provides a unified collection of mathematical formulas and descriptions for machine learning evaluation. The dataset, last updated in April 2026, consolidates performance measures, LIME interpretability measures, and statistical validation tests. Its small size suggests it is a reference sheet rather than a large observational dataset.

Use Cases

• Reference mathematical formulas for model evaluation based on the described performance measures.
• Compare interpretability methods based on the included LIME interpretability measures.
• Select appropriate validation tests based on the described statistical test formulas.

Strengths

• Explicitly includes mathematical formulas for each measure, providing technical precision.
• Unifies three related evaluation domains (performance, interpretability, validation) in one 5.5 KB resource.
• Released under a permissive CC-BY-4.0 license, allowing for broad reuse and modification.

Limitations

• The dataset is very small (5.5 KB), indicating limited scope and likely no observational data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for automated processing.

Provenance

Source

figshare, author Dan Jiang.

Collection Method

Likely compiled from literature and standard references.

Time Range

Creation/update date is 2026-04-21; temporal coverage of the formulas is not specified.

Freshness

Last updated 2026-04-21 17:32:54; freshness should be verified.

Geography

Not applicable; the dataset contains mathematical formulas, not geospatial data.