Seaborn FacetGrid 'hue' changed number of bars

Question

I want to plot facetgrid with seaborn and use 'hue' to identify the number is large or small. However, when I includes 'hue', the number of bars changed.

Generating data

import pandas as pd
import numpy as np
import seaborn as sns

np.random.seed(0)

groups = ('Group 1', 'Group 2')
means = ('Low', 'High', 'Mid')
index =  pd.MultiIndex.from_product(
    [groups, means], 
   names=['Group', 'Mean']
)
values = np.random.randint(low=20, high=100, size=len(index))
data = pd.DataFrame(data={'val': values}, index=index).reset_index()
data['is_positive'] = data['val'] > 75

    Group   Mean    val is_positive
0   Group 1 Low     64  False
1   Group 1 High    67  False
2   Group 1 Mid     84  True
3   Group 2 Low     87  True
4   Group 2 High    87  True
5   Group 2 Mid     29  False

If I don't use 'hue', it shows all bars. However, the color_palette is not applied.

colors = ["red", "green"]
customPalette = sns.set_palette(sns.color_palette(colors))
g = sns.FacetGrid(data, col="Group", hue='is_positive', height=3, col_wrap=6, palette=customPalette)
g.map(sns.barplot, 'val', 'Mean')

If I use 'hue', the color is correct but it only shows two bars.

Ideal chart will show all bars and apply the color_palette. Please help and explain it. Thanks!

Quang Hoang Quang Hoang · Accepted Answer · 2021-02-16T02:15:40

I think you need to pass order option to barplot:

g = sns.FacetGrid(df, col="Group", height=3, hue='is_positive', 
                  col_wrap=6, palette=customPalette)
g.map(sns.barplot, 'val', 'Mean', order=['Low','Mid','High'])

Output:

Seaborn FacetGrid 'hue' changed number of bars

2 Answers