scikit_posthocs.posthoc_tukey_hsd

scikit_posthocs.posthoc_tukey_hsd(a: list | ndarray | DataFrame, val_col: str | None = None, group_col: str | None = None, sort: bool = True) → DataFrame

Pairwise comparisons with Tukey’s honestly significant difference test, using SciPy’s scipy.stats.tukey_hsd implementation.

Parameters:

  • a (array_like or pandas DataFrame) – Two-dimensional data containing the groups to compare. For a list or array, each row is a group; groups may have different lengths in a list.

  • val_col (str or int, optional) – Column containing observations when a is a DataFrame. Required for DataFrame input.

  • group_col (str or int, optional) – Column containing group labels when a is a DataFrame. Required for DataFrame input.

  • sort (bool, optional) – Sort by group and observation value before comparison. Default is True.

Returns:

result – Pairwise p-values, with group labels as rows and columns and ones on the diagonal.

Return type:

pandas.DataFrame

Notes

Refer to scipy.stats.tukey_hsd for details of the test.

Examples

>>> import scikit_posthocs as sp
>>> x = [[1, 2, 3, 4, 5], [35, 31, 75, 40, 21], [10, 6, 9, 6, 1]]
>>> p_values = sp.posthoc_tukey_hsd(x)
>>> round(float(p_values.loc[1, 2]), 6)
0.000991