Seaborn factor plot hue requires no empty “cells”

Question

import pandas as pd
import seaborn as sns
df = pd.DataFrame({'x1': ['a','a','a','b'],
                   'x2': ['c','d','c','d'],
                   'y': [3,8,15,25]})

I would like to see the relationship between x1,x2, and y on the same plot. I love seaborn's factorplot so I was thinking:

sns.factorplot('x1','y',hue='x2',data=df,kind='point')

Unfortunately, it seems that factorplot throws an error if there is an empty cell in pd.crosstab(df.x1,df.x2) . In particular, the error is:

ValueError: low >= high

Not sure why factorplot would be unable to do this -- couldn't the empty intersection of x1,x2 just not be plotted?

For now I'm using row instead of hue for x2 , which splits the plot into rows. Is there a workaround to get the same behavior of hue ? Is there something important that I don't understand about why this error happens?

Answer 1

这应该在开发版本（即0.4.dev ）中0.4.dev ，尝试使用pip install git+git://github.com/mwaskom/seaborn.git#egg=seaborn 。