How can I calculate and place number of observations per seaborn boxplot split by hue?

Question

I've been developing a boxplot with seaborn and have been trying to add an indication of how many observation are in each boxplot.

This article was great to start and this answer has been really helpful and has gotten me most of the way, however, I haven't understood Paul H's comment on how I can use 'hue' to further split up observation calculation and printing (According to Paul it's a basic case of 'passing the hue column by itself' in the example they specify which is different).

I've added my hue ('Condition') within the groupby, however, doesn't seem to have done anything.

#set palette for different hues    
my_pal = {"Year A": "r", "Year B": "B"}

#setup graph    
plt.figure(figsize=(16, 10))
    ax = sns.boxplot(x='variable', y="value", hue="Condition", showmeans=True, data=df, palette=my_pal, meanprops={"marker":"s","markerfacecolor":"white", "markeredgecolor":"black"})
    plt.ylabel("Temperature (\xb0C)")
    plt.xlabel("Zone")

medians = df.groupby(['variable','Condition'])['value'].median().values
nobs = df.groupby(['variable','Condition']).size()
nobs = [str(x) for x in nobs.tolist()]
nobs = ["n: " + i for i in nobs]

pos = range(len(nobs))
for tick,label in zip(pos,ax.get_xticklabels()):
    ax.text(pos[tick], medians[tick] + 0.03, nobs[tick],
    horizontalalignment='center', size='x-small', color='b', weight='semibold')

Thanks!

Answer 1

I had the same problem. I found that , xticklabels are only about x-groups and but no hue-groups in each x-group.

So a simple and brutal force solution is that where you organize your output string, you can output hue-groups in one place instead of one at a time. All the numbers can be written on the top or bottom of the figure, i.e., a fixed y location, not necessary to put on where the medians are.