如何为Matplotlib箱线图的每个类别添加记录数量？

Great question! Adding sample size labels makes your boxplot way more informative—here's a straightforward way to extend your existing code to include them:

Step 1: Calculate record counts per stage #

First, we’ll get the number of observations for each stage, making sure the order matches how stages appear in your boxplot:

# Get the count of records for each stage (order matches boxplot categories)
stage_counts = df.groupby('stage').size().values

Step 2: Draw the boxplot and capture the axis object #

Modify your original code to store the plot’s axis (ax)—this lets us add text annotations later:

# Create your boxplot and save the axis reference
ax = df.assign(index=df.groupby('stage').cumcount())\
       .pivot('index', 'stage', 'days')\
       .plot(kind='box', sym='bD', showmeans=True)

Step 3: Add count labels to each box #

Loop through each stage’s count and place a label above its corresponding box. We’ll position labels near the top of the plot (adjust the y-position if you want them closer to the boxes):

# Add "n=X" labels above each box
for idx, count in enumerate(stage_counts):
    ax.text(
        x=idx,
        y=ax.get_ylim()[1] * 0.95,  # 95% of the y-axis maximum value
        s=f'n={count}',
        ha='center',  # Center the label horizontally over the box
        fontsize=10
    )

Full Combined Code #

Here’s everything put together, with optional plot tweaks:

import pandas as pd
import matplotlib.pyplot as plt

# Your existing data (replace with your actual DataFrame)
# df = pd.read_csv("your_data.csv")

# Calculate stage record counts
stage_counts = df.groupby('stage').size().values

# Generate boxplot with axis reference
ax = df.assign(index=df.groupby('stage').cumcount())\
       .pivot('index', 'stage', 'days')\
       .plot(kind='box', sym='bD', showmeans=True)

# Add sample count labels
for idx, count in enumerate(stage_counts):
    ax.text(idx, ax.get_ylim()[1] * 0.95, f'n={count}', ha='center', fontsize=10)

# Optional: Customize plot labels
ax.set_title('Days by Stage (with Sample Counts)')
ax.set_ylabel('Number of Days')

plt.show()

Quick Notes #

• If you prefer labels right above each box instead of the plot’s top, you can extract the boxplot’s statistical data to target each box’s upper quartile position. The fixed y-position above works for most use cases though.
• The ha='center' parameter ensures each label is perfectly aligned over its box.

内容的提问来源于stack exchange，提问作者Edamame