如何在分组柱状图（grouped bar plot）中添加误差棒

我尝试在分组柱状图中添加误差棒，但目前仅能为所有柱子添加一组，希望为每个ATPS分组内不同Time类别的柱子添加对应误差棒。

数据集示例 #

当前绘图代码 #

import matplotlib.pyplot as plt
import numpy as np

# 假设数据
ATPS_groups = ['A', 'B']
Time_categories = ['0h', '24h']
means = [[5.2, 6.1], [4.8, 5.5]]
stds = [[0.3, 0.4], [0.2, 0.3]]

x = np.arange(len(ATPS_groups))
width = 0.35

fig, ax = plt.subplots()
rects1 = ax.bar(x - width/2, [m[0] for m in means], width, label='0h')
rects2 = ax.bar(x + width/2, [m[1] for m in means], width, label='24h')

# 当前错误的误差棒添加方式（仅添加了一组）
ax.errorbar(x, [m[0] for m in means], yerr=[s[0] for s in stds], fmt='none', c='black')

ax.set_xticks(x)
ax.set_xticklabels(ATPS_groups)
ax.legend()
plt.show()

期望效果 #

每个ATPS分组内，0h和24h的柱子分别对应各自的误差棒（标注于柱子顶部中心位置）

核心是让误差棒的x坐标与对应类别的柱子位置完全对齐，以下是修正后的实现：

Matplotlib原生实现 #

import matplotlib.pyplot as plt
import numpy as np

# 数据集
ATPS_groups = ['A', 'B']
Time_categories = ['0h', '24h']
means = [[5.2, 6.1], [4.8, 5.5]]
stds = [[0.3, 0.4], [0.2, 0.3]]

x = np.arange(len(ATPS_groups))
width = 0.35

fig, ax = plt.subplots()
# 绘制两类柱子
rects1 = ax.bar(x - width/2, [m[0] for m in means], width, label='0h')
rects2 = ax.bar(x + width/2, [m[1] for m in means], width, label='24h')

# 为0h柱子添加对应误差棒，x坐标匹配柱子中心
ax.errorbar(x - width/2, [m[0] for m in means], yerr=[s[0] for s in stds], 
            fmt='none', c='black', capsize=5)
# 为24h柱子添加对应误差棒，x坐标匹配柱子中心
ax.errorbar(x + width/2, [m[1] for m in means], yerr=[s[1] for s in stds], 
            fmt='none', c='black', capsize=5)

ax.set_xticks(x)
ax.set_xticklabels(ATPS_groups)
ax.legend()
plt.show()

Pandas结合Matplotlib实现 #

如果用Pandas处理数据集，思路完全一致：

import pandas as pd
import matplotlib.pyplot as plt

# 构造DataFrame
df = pd.DataFrame({
    'ATPS': ['A', 'A', 'B', 'B'],
    'Time': ['0h', '24h', '0h', '24h'],
    'Mean': [5.2, 6.1, 4.8, 5.5],
    'Std': [0.3, 0.4, 0.2, 0.3]
})

pivot_df = df.pivot(index='ATPS', columns='Time', values=['Mean', 'Std'])
x = np.arange(len(pivot_df.index))
width = 0.35

fig, ax = plt.subplots()
rects1 = ax.bar(x - width/2, pivot_df['Mean']['0h'], width, label='0h')
rects2 = ax.bar(x + width/2, pivot_df['Mean']['24h'], width, label='24h')

# 对应添加误差棒
ax.errorbar(x - width/2, pivot_df['Mean']['0h'], yerr=pivot_df['Std']['0h'], 
            fmt='none', c='black', capsize=5)
ax.errorbar(x + width/2, pivot_df['Mean']['24h'], yerr=pivot_df['Std']['24h'], 
            fmt='none', c='black', capsize=5)

ax.set_xticks(x)
ax.set_xticklabels(pivot_df.index)
ax.legend()
plt.show()

关键说明 #

• 分组柱状图中不同Time类别的柱子x坐标为x ± width/2，误差棒必须使用完全相同的x坐标才能对齐
• capsize参数可控制误差棒两端短横线的长度，提升图表可读性

内容的提问来源于stack exchange，提问作者David Moldes