v2.0.3

What's Changed

• Add support for seeding RNG used for random sampling (#59). by @lebedov in #61

New Contributors

• @lebedov made their first contribution in #61

Full Changelog: v2.0.2...v2.0.3

https://pypi.org/project/gap-stat/#files

v2.0.2

What's Changed

• Update Example.ipynb to support scikit-learn 0.24 by @alinaselega in #51
• Replaced gap_df.append by pd.concat by @psads-git in #58

New Contributors

• @alinaselega made their first contribution in #51
• @psads-git made their first contribution in #58

Full Changelog: v2.0.1...v2.0.2

Ensure passing of numpy arrays to Rust

🐛 Bug fix 🛠️

In v2.0.0, passing a pandas.DataFrame with parallel_backend='rust' would result in TypeError, this patch ensures any dataframe is converted to np.ndarray before hand.

v2.0.0

Disconnect rust & use github actions (#41)
Rust as optional install (#42)
Fix workflows (#43)
Return standard errors of the Gap statistic (#39)
Generate ref data uniformly over feature range (#40)
Pass n_refs and n_iter to rust impl (#46)
Fix Rust ref data generation (#47)
Add warning about optional rust feature (#47)

v2.0.0-rc2

• Pass n_refs and n_iter to rust impl (#46)
• Fix Rust ref data generation (#47)
• Add warning about optional rust feature (#47)

v2.0.0-rc1

Improvements and Fixes:

• Return standard errors of the Gap statistic (#39)
• Generate ref data uniformly over feature range (#40)

Rust as optional install:

• Disconnect rust & use github actions (#41)
• Rust as optional install (#42)
• Fix workflows (#43)

v1.7.1

• Optimizations to Rust backend. #35

v1.7.0

• Change license from BSD-3 to dual Unlicense / MIT (#30)
• Add OSX to Azure Pipelines building (#31)
• Multiple upgrade to gap calculation, n cluster calc, docs, etc (#28)
• Fix passing of null data to final calculations rather than the observed dataset (#34)

Release Candidate 1.7.0.rc1

• Change license from BSD-3 to dual Unlicense / MIT (#30)
• Add OSX to Azure Pipelines building (#31)
• Multiple upgrade to gap calculation, n cluster calc, docs, etc (#28)
• Fix passing of null data to final calculations rather than the observed dataset (#34)

Report reference distribution standard deviations

• OptimalK object's dataframe now has an additional column with the standard deviations of the reference distributions (c61ee92)