Minor Bug in sign_plot leads to all insignificant plot when using DataFrame

[feiden]

Describe the bug
A call by reference in the sign_plot method leads to a bad plot if using a DataFrame.

To Reproduce
Minimal example:

import statsmodels.api as sa
import scikit_posthocs as sp
df = sa.datasets.get_rdataset('iris').data
post_hoc = sp.posthoc_ttest(df, val_col='Sepal.Width', group_col='Species', p_adjust='holm')
fig = sp.sign_plot(post_hoc, flat=False)[0].get_figure()

Solution
Change xc = df.values to xc = df.values.copy() in line 195 in plotting.py

Last Words
Thank you for this nice open source code :)

[maximtrp]

Hi! Thank you for reporting, but there is no bug. Please look at line 194. DataFrames are being copied now, so no need for any other changes.

[feiden]

Hi Maxim, yes the dataframe is being copied as df, but xc is not copied from df. Later df (lines 224-227) is being changed based on the values of xc, but when changing df, xc changes too, leading to an error. You should be able to see that in the code example I provided, no? Maybe this error only occurs with certain versions, idk?

[feiden]

Maybe I should note, that it does not throw an error, but rather the values get incorrect

[maximtrp]

Sorry, did not inspect it carefully. Released the update.