v0.6.9

What's Changed

• add expected groups benchmark by @Illviljan in #218
• Use np.digitize instead of pd.cut by @Illviljan in #217
• Don't use deprecated distutils module. by @sebastic in #221
• Make sure to pass through range indexes by @dcherian in #225
• Add more type hints by @Illviljan in #219
• Micro-optimize: Use ._indexes instead of xindexes. by @dcherian in #226
• Small cleanup in factorize by @dcherian in #220

New Contributors

• @sebastic made their first contribution in #221

Full Changelog: v0.6.8...v0.6.9

v0.6.8

What's Changed

Mostly bugfixes and a small doc improvement

• Recognize grouping by IntervalIndex as binning by @dcherian in #205
• docs: Add introduction to flox by @dcherian in #206
• Propagate multi-indexes properly by @TomNicholas in #216

New Contributors

• @TomNicholas made their first contribution in #216

Full Changelog: v0.6.7...v0.6.8

v0.6.7

What's Changed

• More accurate coverage by @dcherian in #62
• 32bit: More intp fixes by @dcherian in #202

Full Changelog: v0.6.6...v0.6.7

v0.6.6

What's Changed

• Silence more warnings, engine='flox' by @dcherian in #198
• 32bit support: int64 to intp for count by @dcherian in #201

Full Changelog: v0.6.5...v0.6.6

v0.6.5

This release brings a major bugfix for grouping by multiple variables.

What's Changed

• Fix race condition in factorize by @dcherian in #196
• Suppress warnings by @dcherian in #197

Full Changelog: v0.6.4...v0.6.5

v0.6.4

What's Changed

• Avoid broadcasting by variables against each other by @dcherian in #186
• Update visualizations & docs by @dcherian in #189
• Check for exact alignment by @dcherian in #192
• Consolidate validation of expected_groups by @dcherian in #193
• Add windows CI by @Illviljan in #151

Full Changelog: v0.6.3...v0.6.4

v0.6.3

What's Changed

Performance updates:

• Fix blockwise sort optimization by @dcherian in #181
• Use blockwise to extract final result by @dcherian in #182

Full Changelog: v0.6.2...v0.6.3

v0.6.2

What's Changed

Breaking Changes

Bump min numpy_groupies to >= 0.9.19

Enhancements

• Speed up var, std by using using numpy_groupies sum_of_squares by @dcherian in #135. For this we bump min numpy_groupies to >= 0.9.19
• Support reindexing in simple_combine by @dcherian in #177. This is a big improvement when distributed's worker-saturation parameter is inf. If 1.2 it doesn't really change much.

Bug fixes

• Force reindex to be bool always by @dcherian in #176

Full Changelog: v0.6.1...v0.6.2

v0.6.1

This release fixes a really bad performance regression with method="cohorts" and much improved cohorts detection

What's Changed

• Performance improvements for cohorts detection by @dcherian in #172
• Major fix to subset_to_blocks by @dcherian in #173

Full Changelog: v0.6.0...v0.6.1

v0.6.0

What's Changed

Enhancements

• Better dtype support:
  • Add a dtype check for numpy arrays in assert_equal by @Illviljan in #158
  • Allow specifying output dtype by @dcherian in #131
• Substantially more efficient cohorts strategy by @dcherian in #165

Deprecations and Removals

• Deprecate resample_reduce by @dcherian in #169
• Remove split_out by @dcherian in #170

Full Changelog: v0.5.10...v0.6.0