Skip to content

Releases: pjaselin/Cubist

v0.1.4

30 May 02:03
35471c1
Compare
Choose a tag to compare

What's Changed

  • Feature/release patch by @pjaselin in #140
  • adding pattern matching uploads/downloads of artifacts by @pjaselin in #142
  • testing to verify Cubist can handle nan cells by @pjaselin in #143
  • added check to handle pandas applymap -> map change by @pjaselin in #147

Thanks to @ramongss for bringing this up!

Full Changelog: v0.1.3...v0.1.4

v0.1.3

21 Mar 02:34
4f21a5c
Compare
Choose a tag to compare

Increased test coverage and bug fixes

  • Handled presence of "sample" name in input columns to match R code
  • Moved all metadata to pyproject.toml file and removed other config files
  • Removed unused sections of code
  • Better parameter validation sequences and more consistent messages
  • Started testing full datasets for completion (need to test against reference models for benchmarking tests)
  • Added separate workflow step for signing and uploading released objects

v0.1.2

08 Jan 00:35
904b38b
Compare
Choose a tag to compare

Bug fix:

  • Found the R format function was not translated correctly while adding tests for better code coverage. Found that np.format_float_positional with unique=False, fractional=False, trim="-" achieves the same output.

Code quality:

  • Using Ruff for linting
  • Added pre-commit hooks
  • Added code coverage

v0.1.1

19 Nov 05:20
f5451f5
Compare
Choose a tag to compare
  • Enabling support for Cython 3.11
  • Caught up to R package changes in C code where possible (excluding usage of size_t)
  • Updating package for cleaner code (flake8),
  • Cleaned up readme and some docstrings
  • Closing #121

v0.1.0

07 Mar 17:40
a3b8a53
Compare
Choose a tag to compare

Breaking changes:

  • Made the API match the discussed changes in this PR for the R package: topepo/Cubist#40
  • Whether a model is composite is determined by the neighbors parameter and whether Cubist automatically chooses whether to use a composite model is determined by the auto parameter.

Other:

  • Docstring and README updates
  • More comprehensive and better written tests
  • Made all internal functions private
  • General file cleanups

v0.0.16

29 Dec 03:59
b65c49c
Compare
Choose a tag to compare
  • Better error handling
  • output label will be the Pandas series label if available

v0.0.15

03 Dec 01:39
0f4da25
Compare
Choose a tag to compare

Enabled cross validation and whether to use composite models, more sklearn-like default parameters

v0.0.14

06 Nov 17:25
3e667df
Compare
Choose a tag to compare

Adding Windows support per #70

v0.0.13

26 Sep 02:59
72ff319
Compare
Choose a tag to compare
Merge pull request #69 from pjaselin/master

release v0.0.13

v0.0.11

04 Aug 22:33
Compare
Choose a tag to compare
cleaned up cubist class