v0.5.5

This is a minor release with several bugfixes, and major updates to the internal structure and sphinx documentation.

What's Changed

• Fix penalty for LogisticRegression by @raphaelvallat in #403
• Switch to modern python packaging by @getzze in #406
• Remove call to sns.despine by @raphaelvallat in #410
• Updated deprecated function by @sjg2203 in #414
• Add errstate(divide="ignore") in Bayes Factor calculation by @raphaelvallat in #415
• Remove inplace on single column by @raphaelvallat in #423
• Fix RBC sign in mwu by @raphaelvallat in #424
• Overhaul documentation (pydata_sphinx_theme) by @yann1cks in #432
• Release 0.5.5 by @raphaelvallat in #436

New Contributors

• @getzze made their first contribution in #406
• @sjg2203 made their first contribution in #414
• @yann1cks made their first contribution in #432

Full Changelog: v0.5.4...v0.5.5

v0.5.4

This is a minor release with several bugfixes and no new features. The new version is tested for Python 3.8-3.11 (but should also work with Python 3.12).

This release requires pandas≥1.5. We recommend scipy≥1.11.0.

What's Changed

• Minor typo fix in docs by @musicinmybrain in #329
• clip r values by @remrama in #342
• fix: deprecated parameter by @bitsnaps in #341
• hotfix: CI crash in test_power_chi2 [WIP] by @raphaelvallat in #344
• hotfix: plot_rm_corr crash with specific column names by @remrama in #351
• Add check for noncentrality parameters. by @agkphysics in #347
• Use pyupgrade by @raphaelvallat in #364
• fix groupby.mean for only numeric values by @jajcayn in #363
• Function test fails for np.mean by @gedeck in #380
• Fix in flatten_list for Python 3.12 by @raphaelvallat in #370
• corr(): fix CI95% column name in returned dataframe by @kraktus in #382
• Replace None in dataset to fix unit tests by @raphaelvallat in #388
• Remove outdated + bump pandas 1.5 by @raphaelvallat in #389
• Fix doctests by @raphaelvallat in #390
• Fix warnings by @raphaelvallat in #391
• Remove non-centrality check (solved in scipy 1.11) by @raphaelvallat in #392
• Use numeric_only=True in DataFrame.corr() and cov() by @raphaelvallat in #393
• Add numeric_only=True in remaining pandas functions by @raphaelvallat in #396
• Release 0.5.4 by @raphaelvallat in #397

New Contributors

• @musicinmybrain made their first contribution in #329
• @bitsnaps made their first contribution in #341
• @agkphysics made their first contribution in #347
• @jajcayn made their first contribution in #363
• @kraktus made their first contribution in #382

Full Changelog: v0.5.3...v0.5.4

v0.5.3

This is a minor release with a few bugfixes, several improvements and one new function/pandas.DataFrame method. Read the changelog at https://pingouin-stats.org/changelog.html

What's Changed

• Fix numerical stability issue in multivariate_normality by @gkanwar in #292
• Add new function for pairwise T-tests between columns of a dataframe (pingouin.ptests) by @raphaelvallat in #291
• Handle single-sample comparsion in pairwise_test by @George3d6 in #299
• Change TestRegression class test methods to fix victim flakiness by @blazyy in #303
• Add aesthetic flexibility to plot_rm_corr by @remrama in #312
• Update distribution.py by @ALL-SPACE-Rob in #310
• Plotting seaborn.FacetGrid compatibility by @remrama in #314
• Use scikit-learn>=1.1.2 by @raphaelvallat in #300
• Plot shift documentation PR by @turkalpmd in #320
• Fix pandas warning by @raphaelvallat in #323
• Deal with small sample size in pingouin.normality when using long-format by @raphaelvallat in #324
• Renamed 'r' with 'pointbiserialr' in convert_effsize by @raphaelvallat in #325
• Exact calculation of effect sizes in pairwise_tukey and pairwise_gameshowell by @raphaelvallat in #328

New Contributors

• @gkanwar made their first contribution in #292
• @George3d6 made their first contribution in #299
• @blazyy made their first contribution in #303
• @remrama made their first contribution in #312
• @ALL-SPACE-Rob made their first contribution in #310
• @turkalpmd made their first contribution in #320

v0.5.2

Bugfixes

a. The eta-squared (n2) effect size was not properly calculated in one-way and two-way repeated measures ANOVAs. Specifically, Pingouin followed the same behavior as JASP, i.e. the eta-squared was the same as the partial eta-squared. However, as explained in #251, this behavior is not valid. In one-way ANOVA design, the eta-squared should be equal to the generalized eta-squared. As of March 2022, this bug is also present in JASP. We have therefore updated the unit tests to use JAMOVI instead.

Please double check any effect sizes previously obtained with the pingouin.rm_anova function!

b. Fixed invalid resampling behavior for bivariate functions in pingouin.compute_bootci when x and y were not paired. #281
c. Fixed bug where confidence (previously ci) was ignored when calculating the bootstrapped confidence intervals in pingouin.plot_shift. #282

Enhancements

a. The pingouin.pairwise_ttests has been renamed to pingouin.pairwise_tests. Non-parametric tests are also supported in this function with the parametric=False argument, and thus the name "ttests" was misleading #209
b. Allow pingouin.bayesfactor_binom to take Beta alternative model. #252
c. Allow keyword arguments for logistic regression in pingouin.mediation_analysis. #245
d. Speed improvements for the Holm and FDR correction in pingouin.multicomp. #271
e. Speed improvements univariate functions in pingouin.compute_bootci (e.g. func="mean" is now vectorized).
f. Rename eta to eta_squared in pingouin.power_anova andpingouin.power_rm_anova to avoid any confusion. #280
g. Add support for DataMatrix objects. #286
h. Use black for code formatting.

v0.5.1

Pingouin 0.5.1

This is a minor release, with several bugfixes and improvements. This release is compatible with SciPy 1.8 and Pandas 1.4.

Bugfixes

• Added support for SciPy 1.8 and Pandas 1.4. #234
• Fixed bug where pingouin.rm_anova() and pingouin.mixed_anova() changed the dtypes of categorical columns in-place #224

Enhancements

• Faster implementation of pingouin.gzscore(), adding all options available in zscore: axis, ddof and nan_policy. Warning: this function is deprecated and will be removed in the next version of Pingouin (use scipy.stats.gzscore() instead). #210.
• Replace use of statsmodels’ studentized range distribution functions with more SciPy’s more accurate scipy.stats.studentized_range(). #229.
• Add support for optional keywords argument in the pingouin.homoscedasticity() function #218
• Add support for the Jarque-Bera test in pingouin.normality() #216.

Lastly, we have also deprecated the Gitter forum in favor of GitHub Discussions. Please use Discussions to ask questions, share ideas / tips and engage with the Pingouin community!

v0.5.0

This is a major release with several important bugfixes. We recommend all users to upgrade to this new version.

See the full changelog at: https://pingouin-stats.org/changelog.html#v0-5-0-october-2021

v0.4.0

This is a major release with an important upgrade of the dependencies (requires Python 3.7+ and SciPy 1.7+), several enhancements in existing function and a new function to test the equality of covariance matrices (pingouin.box_m). We recommend all users to upgrade to the latest version of Pingouin.

See the full changelog at: https://pingouin-stats.org/changelog.html#v0-4-0-august-2021

v0.3.12

This release fixes a critical error in pingouin.partial_corr: the number of covariates was not taken into account when calculating the degrees of freedom of the partial correlation, thus leading to incorrect results (except for the correlation coefficient which remained unaffected). For more details, please see #171.

For the full changelog, please see https://pingouin-stats.org/changelog.html

v0.3.11

This is a minor release with several bug fixes in pingouin.corr. The full changelog can be found here.

v0.3.10

This release fixes an error in the calculation of the p-values in the pg.pairwise_tukey() and pg.pairwise_gameshowell() functions (#156). Old versions of Pingouin used an incorrect algorithm for the studentized range approximation, which resulted in (slightly) incorrect p-values. In most cases, the error did not seem to affect the significance of the p-values. The new version of Pingouin uses statsmodels to estimate the p-values.