如何在Python中实现类似R的多图合并（grid.arrange/facet_wrap）功能？

Great question! In Python, you have two go-to approaches to replicate the facetting/arranging workflow you’re familiar with from R. Let’s walk through both—one that’s quick and high-level (using seaborn) and another for full manual control (using matplotlib).

Option 1: Use Seaborn's FacetGrid (Most Like R's facet_wrap) #

Seaborn is built on matplotlib and has a FacetGrid class that’s perfect for creating grouped subplots just like facet_wrap. It handles the layout and subplot mapping automatically, so you can focus on your density plots.

Example Code: #

import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

# Create sample data (replace with your actual dataset)
df = pd.DataFrame({
    "weekday": pd.Categorical(
        ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"],
        categories=["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"],
        ordered=True  # Ensures weekdays stay in correct order
    ),
    "category": ["A", "B"] * 350,  # Your categorical variable
    "value": np.random.normal(loc=0, scale=1, size=700)  # Your continuous variable
})

# Set up the facet grid
g = sns.FacetGrid(
    df,
    col="weekday",  # Split subplots by weekday
    col_wrap=3,     # Number of subplots per row
    height=3,       # Height of each subplot
    aspect=1        # Width-to-height ratio of each subplot
)

# Map the density plot to each subplot
g.map(
    sns.kdeplot,
    "value",
    hue="category",  # Differentiate by your categorical variable
    fill=True,       # Fill under the density curve
    common_norm=False  # Keep densities per subplot independent
)

# Add a shared legend and clean up the layout
g.add_legend(title="Category")
plt.tight_layout()
plt.show()

Key Notes: #

• col_wrap lets you control how many subplots fit in each row, just like facet_wrap’s ncol parameter.
• Using pd.Categorical with ordered=True ensures your weekdays appear in the correct order (not alphabetical).
• common_norm=False ensures each subplot’s density is scaled independently, which is often useful for comparing distributions across days.

Option 2: Manual Layout with Matplotlib subplots (Like R's grid.arrange) #

If you want full control over every aspect of the subplot layout, use matplotlib’s subplots function. This is analogous to manually arranging plots with grid.arrange in R.

Example Code: #

import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
import numpy as np

# Reuse the same sample data (or replace with yours)
df = pd.DataFrame({
    "weekday": pd.Categorical(
        ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"],
        categories=["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"],
        ordered=True
    ),
    "category": ["A", "B"] * 350,
    "value": np.random.normal(loc=0, scale=1, size=700)
})

# Create a 3x3 grid of subplots (fits 7 weekdays with 2 empty spots)
fig, axes = plt.subplots(nrows=3, ncols=3, figsize=(12, 9))
axes = axes.flatten()  # Convert 2D array of axes to 1D for easier looping

# Loop through each weekday and plot its density
for idx, day in enumerate(df["weekday"].cat.categories):
    # Filter data for the current weekday
    day_data = df[df["weekday"] == day]
    
    # Plot density curves for each category
    sns.kdeplot(
        data=day_data,
        x="value",
        hue="category",
        fill=True,
        common_norm=False,
        ax=axes[idx]  # Assign plot to the correct subplot axis
    )
    
    # Customize subplot titles and labels
    axes[idx].set_title(f"Density: {day}")
    axes[idx].set_xlabel("Continuous Variable")
    axes[idx].set_ylabel("Density")

# Hide any empty subplots (since 3x3=9, but we only have 7 days)
for ax in axes[len(df["weekday"].cat.categories):]:
    ax.axis("off")

# Clean up spacing between subplots
plt.tight_layout()
plt.show()

Key Notes: #

• figsize lets you control the overall size of the combined plot.
• Flattening the axes array makes it easy to loop through each subplot position.
• You can fully customize each subplot (titles, labels, colors) individually if needed.

Which Should You Choose? #

• Use Seaborn FacetGrid if you want a quick, clean solution that handles most of the layout work for you (perfect for replicating facet_wrap).
• Use Matplotlib subplots if you need granular control over every part of the plot layout (like grid.arrange).

内容的提问来源于stack exchange，提问作者SRVFan