如何根据熊猫中的给定键更改背景单元格颜色？

Question

我有以下数据框：

df = pd.DataFrame({"A":["foo", "foo", "foo", "bar"],  
                   "B":["A","A","B","A"], 
                   "C":[0,3,1,1]})

如何更改 A 列和 B 列的单元格颜色，按其值分组。我的意思是，这将是理想的输出：

可能是这样的：

df.groupby(by=['A', 'B']).style.change_background()

由于真正的数据框有数百行，我会对自动分配颜色感兴趣。

Answer 1

from matplotlib.pyplot import cm 
from matplotlib.colors import to_hex

# convert groups to indices    
g = pd.Categorical(df.A + df.B).codes

# generate a list hex colors
color = cm.rainbow(g / g.max())
hex_color = [to_hex(col) for col in color]
print('generate colors: ', hex_color)

# apply the colors to style for columns A and B
df.style.apply(
    lambda col: ['background-color: %s' % (hex_color[i]) for i in range(col.size)], 
    subset=['A', 'B']
)

Antonvbr 更新：（受此解决方案的启发，我也添加了一个使用 seaborn 的解决方案。

import seaborn as sns
import pandas as pd

df = pd.DataFrame({"A":["foo", "foo", "foo", "bar"],  
                   "B":["A","A","B","A"], 
                   "C":[0,3,1,1]})

g = pd.Categorical(df.A + df.B).codes # convert groups to indices  
n = np.unique(g).size
palette = sns.color_palette("hls", n).as_hex()

# apply the colors to style for columns A and B
df.style.apply(
    lambda x: ['background-color: {}'.format(palette[i]) for i in g], 
    subset=['A', 'B']
)