Roadmap for release 0.5.2

[raphaelvallat]

The following issues should be addressed:

• bayesfactor_binom should allow H1 distribution to be specified #249
• Mediation Analysis: Pass on arguments to logistic_regression #244
• pairwise_nonparametric() #209
• [Bugfix] compute_bootci should resample x and y separately when paired=False #276
• Use scipy.stats.bootstrap in pingouin.compute_bootci? #189
• LogisticRegression fails when X is a single predictor #272
• pg.multicomp() uses a lot of memory, crashes the system #270
• power_anova eta parameter is actually eta-squared #268
• Do eta squared and partial eta squared coincide in one-way repeated measures anova? #251
• shift_plt ci argument gets ignored #259

The following PR should be merged:

• Rename pairwise_ttests to pairwise_tests #246
• Allow bayesfactor_binom to take Beta alternative model (#249) #252
• optional arguments for logistic_regression() in mediation_analysis() #245
• Speed improvements for pg.multicomp #271
• Fix eta-squared calculation in repeated measures ANOVA design #254
• Renamed eta to eta-squared in power functions #280
• Add support for DataMatrix objects #286

The following dependencies should be updated:

• Scikit-learn <1.1.0 (to avoid bug in logistic regression)

The following deprecation should be made:

[raphaelvallat]

Pingouin 0.5.2 is now ready! I'll release tomorrow.

[raphaelvallat]

Released: https://github.com/raphaelvallat/pingouin/releases/tag/v0.5.2