v0.9.14

What's Changed

• Support first, last with datetime, timedelta by @dcherian in #402

Full Changelog: v0.9.13...v0.9.14

v0.9.13

What's Changed

• Faster subsetting for cohorts by @dcherian in #397

Full Changelog: v0.9.12...v0.9.13

v0.9.12

What's Changed

• Avoid rechunking when preferred_method="blockwise" by @dcherian in #394
• Fix default int on windows, numpy<2 by @dcherian in #395

Full Changelog: v0.9.11...v0.9.12

v0.9.11

What's Changed

• Drop python 3.9, use ruff by @dcherian in #392
• Preserve dtype better when specified. by @dcherian in #389

Full Changelog: v0.9.10...v0.9.11

v0.9.10

Fixes a couple of bugs.

What's Changed

• Fix first, last again by @dcherian in #381
• Avoid explicit np.nan, np.inf by @dcherian in #383
• Fix bug with NaNs in by and method='blockwise' by @dcherian in #384
• Expand groupby_reduce property tests by @dcherian in #385

Full Changelog: v0.9.9...v0.9.10

v0.9.9

Faster cohorts detection, property tests, and a beta quality implementation of scans (nancumsum, ffill, bfill)

What's Changed

• Property tests with hypothesis by @dcherian in #348
• Add scans by @dcherian in #370
• Optimize for-loop merging of cohorts. by @dcherian in #378

Full Changelog: v0.9.8...v0.9.9

v0.9.8

What's Changed

• Support method="blockwise" with Cubed. by @tomwhite in #357

Full Changelog: v0.9.7...v0.9.8

v0.9.7

Some performance improvements, Numpy 2 compat, and initial support for Cubed!

What's Changed

• Initial minimal working Cubed example for "map-reduce" by @tomwhite in #352
• Fix benchmarks by @dcherian in #358
• Optimize bitmask finding for chunk size 1 and single chunk cases by @dcherian in #360
• Add cubed notebook for hourly climatology example using "map-reduce" method by @tomwhite in #356
• Manually fuse reindexing intermediates with blockwise reduction for cohorts. by @dcherian in #300
• Use threadpool for finding labels in chunk by @dcherian in #327
• Optimize min_count when expected_groups is not provided. by @dcherian in #236
• import normalize_axis_index from numpy.lib on numpy>=2 by @keewis in #364

New Contributors

• @tomwhite made their first contribution in #352

Full Changelog: v0.9.6...v0.9.7

v0.9.6

What's Changed

• Speed up method=None heuristics by avoiding dot product as much as possible by @dcherian in #347

Full Changelog: v0.9.5...v0.9.6

v0.9.5

What's Changed

• Fix nanlen with strings by @dcherian in #344

Full Changelog: v0.9.4...v0.9.5