如何用Matplotlib绘制各年份堆叠国家不同的堆叠柱状图？

Hey there! I totally get the frustration of dealing with duplicate legends and clunky loop-based plotting for stacked bars. Let's simplify this using pandas' built-in plotting tools—they'll handle stacking, color consistency, and clean legends in one efficient go.

Step 1: Reshape Your Data for Stacked Plotting #

First, we need to convert your long-format DataFrame into a wide format where each year is a row, each country is a column, and values are the total counts. This makes stacked bar plotting trivial:

import pandas as pd
import matplotlib.pyplot as plt

# Your original dataset
data = {
    'year': [2010, 2010, 2011, 2011, 2012, 2012, 2013, 2013],
    'country': ['USA', 'CHIN', 'USA', 'JAPN', 'KORR', 'USA', 'CHIN', 'USA'],
    'total': [10, 12, 8, 12, 7, 10, 9, 13]
}
df = pd.DataFrame(data)

# Reshape to wide format (fill missing country-year pairs with 0)
wide_df = df.pivot(index='year', columns='country', values='total').fillna(0)

Step 2: Assign Unique Colors to Countries #

Create a color mapping dictionary to lock in consistent colors for each country—this ensures your legend and plot segments match perfectly:

country_colors = {
    'USA': '#1f77b4',    # Blue
    'CHIN': '#ff7f0e',   # Orange
    'JAPN': '#2ca02c',   # Green
    'KORR': '#d62728'    # Red
}

Step 3: Plot the Stacked Bar Chart #

Use pandas' plot() method with kind='bar' and stacked=True. Pandas will automatically handle stacking, color assignment, and legend creation—no loops required:

# Generate the stacked bar chart
ax = wide_df.plot(kind='bar', stacked=True, color=country_colors, figsize=(10, 6))

# Customize plot labels and styling
ax.set_title('Total by Year (Stacked by Country)', fontsize=14)
ax.set_xlabel('Year', fontsize=12)
ax.set_ylabel('Total', fontsize=12)
ax.tick_params(axis='both', labelsize=10)

# Adjust legend placement for readability
plt.legend(title='Country', bbox_to_anchor=(1.05, 1), loc='upper left')

plt.tight_layout()
plt.show()

Why This Beats Your Loop Approach #

• No redundant legends: Pandas creates one legend entry per country (not per data point), so you’ll only see 4 entries instead of 8.
• Faster plotting: This vectorized approach skips row-by-row loops, making it way more efficient—especially with larger datasets.
• Cleaner code: You avoid manually tracking bar positions and legend handles, reducing room for errors.

This will give you a polished stacked bar chart where each year has a single bar, split into segments for each country present that year, with consistent colors and a tidy legend.

内容的提问来源于stack exchange，提问作者bill