Post-Hoc test in Python

# Tukey HSD test
# assumptions: equal variances between groups and equal group size
from statsmodels.stats.multicomp import pairwise_tukeyhsd
from statsmodels.stats.multicomp import MultiComparison
mc = MultiComparison(data["value"], data["treatments"])
mcresult = mc.tukeyhsd(0.05)
mcresult.summary()


# After ANOVA if group size is different, we use Tukey-Kramer Post-Hoc test
# If we have Control group and want only these comparisson -> C:A, C:B, C:D, then the appropriate test is Dunnett's Post-Hoc test
# After Welch's ANOVA -> Games-Howell Post-Hoc test, because variances are not equal
# After Friedman we use Wilcoxon Post-Hoc test
# After Kruskal-Wallis we use Dunn Post-Hoc test or Nemenyi Post-Hoc test