PariGrid map_diag does not preserve category ordering

[wmayer7]

PairGrid.map_diag does not preserve the order specified for ordered categorical columns in a pandas DataFrame. (I'm using seaborn 0.9.0 and pandas 0.24.0.)

In the following example, the diagonal axes for the count plots use order a,b,c and d,e,f instead of the category order c,a,b and f,e,d, respectively.

import pandas as pd
import seaborn as sns

X = pd.Categorical(["a", "b", "c", "a"], categories=["c", "a", "b"], ordered=True)
Y = pd.Categorical(["d", "e", "f", "e"], categories=["f", "e", "d"], ordered=True)
df = pd.DataFrame(dict(X=X,Y=Y))
g = sns.PairGrid(df, vars=df.columns)
g.map_diag(sns.countplot)

Could the pandas Series/Group object be passed to the plotting function (instead of converting it to a numpy array) to preserve the full data type information associated with each pandas column?

[mwaskom]

The reason it's not a trivial change is that plt.hist breaks on Series inputs (or at least it used to, that may be fixed in newer versions). In the meantime you can use the ‘order’ kwarg.