Python 使用 Seaborn Relplot 绘制带有列表单元格的数据框

Question

我有一个数据框。两列中的每个单元格都包含一个列表。我想绘制一个 seaborn-relplot ，其中两列为 x 和 y。我正在遇到问题

我的代码:

xdf = pd.DataFrame({'v':[[1,2,10,20],[3,4]],'i':[[5,6,50,60],[7,8]]})
xdf['nor_iv'] = ['True','False']
print(xdf)
xdf = 
                v               i nor_iv
0  [1, 2, 10, 20]  [5, 6, 50, 60]   True
1          [3, 4]          [7, 8]  False

# Seaborn plot
hue_colors = {'True': 'red', 
              'False': 'green'}
sns.relplot(data=xdf,x="v", y="i",hue='nor_iv',
             markers=True,kind='line',
                       palette=hue_colors,height=4, aspect=1.5)
plt.show()

当前输出:

TypeError: unhashable type: 'list'

Answer 1

首先使用xdf = xdf.apply(pd.Series.explode)。请注意，您需要在爆炸后将所需的列从字符串(即 object)转换为数据类型 int。以下三种可视化方法:

如果您想要叠加(无行或列):

import matplotlib.pyplot as plt
import seaborn as sns
xdf = pd.DataFrame({'v':[[1,2,10,20],[3,4]],'i':[[5,6,50,60],[7,8]]})
xdf['nor_iv'] = ['True','False']
xdf = xdf.apply(pd.Series.explode)
xdf['v'] = xdf['v'].astype(int)
xdf['i'] = xdf['v'].astype(int)
sns.relplot(data=xdf,x="v", y="i",hue='nor_iv',
             markers=True,kind='line',
                       palette=hue_colors,height=4, aspect=1.5)
plt.show()

如果你想要列，你可以这样做:

import matplotlib.pyplot as plt
import seaborn as sns
xdf = pd.DataFrame({'v':[[1,2,10,20],[3,4]],'i':[[5,6,50,60],[7,8]]})
xdf['nor_iv'] = ['True','False']
xdf = xdf.apply(pd.Series.explode)
xdf['v'] = xdf['v'].astype(int)
xdf['i'] = xdf['v'].astype(int)
sns.relplot(data=xdf,x="v", y="i",col='nor_iv', hue='nor_iv',
             markers=True,kind='line',
                       palette=hue_colors,height=4, aspect=1.5)
plt.show()

如果你想要行，那么:

import matplotlib.pyplot as plt
import seaborn as sns
xdf = pd.DataFrame({'v':[[1,2,10,20],[3,4]],'i':[[5,6,50,60],[7,8]]})
xdf['nor_iv'] = ['True','False']
xdf = xdf.apply(pd.Series.explode)
xdf['v'] = xdf['v'].astype(int)
xdf['i'] = xdf['v'].astype(int)
sns.relplot(data=xdf,x="v", y="i",row='nor_iv', hue='nor_iv',
             markers=True,kind='line',
                       palette=hue_colors,height=4, aspect=1.5)
plt.show()