为何每个Seaborn柱状图首柱显示0.00%？

问题描述 #

运行基于Pandas、Seaborn、Matplotlib的Python代码绘制以fraud为分组的countplot时，图表整体显示正常，但每个柱状图的第一个柱子上均标注了0.00%。

重现代码 #

dt = pd.DataFrame({'witness':['No', 'No', 'No', 'No', 'No', 'No', 'No', 'Yes', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No'],
              'category':['Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Utility', 'Sedan', 'Sport', 'Sport', 'Sport', 'Sport', 'Sedan', 'Sedan', 'Sport', 'Sport', 'Sport', 'Sedan', 'Sport', 'Sport', 'Sedan', 'Sport', 'Sedan'],
              'make':['Honda', 'Honda', 'Honda', 'Toyota', 'Honda', 'Honda', 'Honda', 'Honda', 'Ford', 'Mazda', 'Honda', 'Ford', 'Ford', 'Ford', 'Ford', 'Chevrolet', 'Pontiac', 'Honda', 'Mazda', 'Chevrolet', 'Mazda', 'Pontiac', 'Mazda', 'Pontiac', 'Honda'],
              'sex':['Female', 'Male', 'Male', 'Male', 'Female', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Female', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male'],
              'fraud':[0, 0, 0, 0,1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0,1]})

fea = ['witness', 'category', 'make', 'sex']

fig, axes =plt.subplots(10,2,figsize=(10,40))
plt.subplots_adjust(wspace=0.9, hspace=0.9)
i=0
total = len(dt['witness']) 

for fea_name in fea:
     ax= axes[i//2, i%2]
     i += 1
     sns.countplot(x=fea_name, data=dt, ax=ax,hue='fraud' )
     ax.set_title(f'Countplot of {fea_name}')
     ax.tick_params(axis='x', rotation=45)

for i in range(len(fea), 10 * 2):
    fig.delaxes(axes[i//2, i%2])
for ax in axes.flat:
    for p in ax.patches:
        percentage = '{:.1f}%'.format(100 * p.get_height() / total)
        x = p.get_x() + p.get_width()/3
        y = p.get_y() + p.get_height()
        ax.annotate(percentage, (x, y))

plt.tight_layout()
plt.show()

问题原因 #

• 代码创建了10×2共20个子图，但实际仅使用4个，剩余16个通过fig.delaxes()删除。
• 标注遍历逻辑用了axes.flat，会包含所有20个子图的轴对象，包括已被删除的轴。
• 已删除的轴没有有效柱子（patches高度为0），但代码仍尝试计算百分比并标注，这些无效标注会被绘制到当前活跃的可见轴上，导致每个子图的第一个柱子出现错误的0.00%。

解决方案 #

方案1：收集并遍历使用过的轴 #

绘制countplot时，将实际使用的轴存入列表，后续仅遍历该列表：

dt = pd.DataFrame({'witness':['No', 'No', 'No', 'No', 'No', 'No', 'No', 'Yes', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No', 'No'],
              'category':['Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Sport', 'Utility', 'Sedan', 'Sport', 'Sport', 'Sport', 'Sport', 'Sedan', 'Sedan', 'Sport', 'Sport', 'Sport', 'Sedan', 'Sport', 'Sport', 'Sedan', 'Sport', 'Sedan'],
              'make':['Honda', 'Honda', 'Honda', 'Toyota', 'Honda', 'Honda', 'Honda', 'Honda', 'Ford', 'Mazda', 'Honda', 'Ford', 'Ford', 'Ford', 'Ford', 'Chevrolet', 'Pontiac', 'Honda', 'Mazda', 'Chevrolet', 'Mazda', 'Pontiac', 'Mazda', 'Pontiac', 'Honda'],
              'sex':['Female', 'Male', 'Male', 'Male', 'Female', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Female', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male', 'Male'],
              'fraud':[0, 0, 0, 0,1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0, 1, 0, 1, 0, 0,1]})

fea = ['witness', 'category', 'make', 'sex']

fig, axes =plt.subplots(10,2,figsize=(10,40))
plt.subplots_adjust(wspace=0.9, hspace=0.9)
i=0
total = len(dt['witness']) 
used_axes = []  # 存储使用过的轴对象

for fea_name in fea:
     ax= axes[i//2, i%2]
     used_axes.append(ax)
     i += 1
     sns.countplot(x=fea_name, data=dt, ax=ax,hue='fraud' )
     ax.set_title(f'Countplot of {fea_name}')
     ax.tick_params(axis='x', rotation=45)

for i in range(len(fea), 10 * 2):
    fig.delaxes(axes[i//2, i%2])

# 仅遍历实际使用过的轴
for ax in used_axes:
    for p in ax.patches:
        percentage = '{:.1f}%'.format(100 * p.get_height() / total)
        x = p.get_x() + p.get_width()/3
        y = p.get_y() + p.get_height()
        ax.annotate(percentage, (x, y))

plt.tight_layout()
plt.show()

方案2：跳过不可见的轴 #

遍历所有轴时，通过ax.get_visible()判断轴是否有效，跳过已删除的轴：

# 替换原代码中的遍历标注部分
for ax in axes.flat:
    if not ax.get_visible():  # 跳过已删除的轴
        continue
    for p in ax.patches:
        percentage = '{:.1f}%'.format(100 * p.get_height() / total)
        x = p.get_x() + p.get_width()/3
        y = p.get_y() + p.get_height()
        ax.annotate(percentage, (x, y))

效果说明 #

修改后，错误的0.00%标注会消失，每个柱子的百分比标注将准确对应其实际计数。

内容的提问来源于stack exchange，提问作者Manish Patel