GeoPandas六边形几何绘制问题：自定义配色与图例展示

Hey there, let's work through this problem step by step — I've dealt with similar GeoPandas plotting glitches before, so let's start with the root cause of the distorted visuals, then nail the custom colors and proper legend.

First: Fix Geometry Topology (The Likely Culprit for Distortion) #

Most of the time, wonky plots with polygons come down to invalid geometry (think self-intersections, overlaps, or malformed shapes). Let's check and fix that first:

# Check for invalid polygons
invalid_rows = df_january[~df_january.geometry.is_valid]
print(f"Found {len(invalid_rows)} invalid geometries")

# Quick fix for common topology issues (buffer(0) works magic here)
df_january['geometry'] = df_january.geometry.buffer(0)

After running this, try plotting again — odds are the distortion will be gone right away. If not, double-check your coordinate reference system (CRS): geographic CRS (like WGS84 lat/lon) can warp hexagons when plotting, so convert to a projected CRS (UTM works great for local areas):

# Auto-detect and convert to UTM projection
df_january = df_january.to_crs(df_january.estimate_utm_crs())

Second: Custom Coloring + Legend for pred_labels (Discrete Categories) #

Since pred_labels are discrete values (like 1, 3), we need to treat them as categories instead of continuous numbers — that's a common mistake that causes weird rendering. Here's how to set up custom colors and a clean legend:

import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap

# Define your custom color mapping (match labels to your preferred shades)
label_color_map = {
    1: '#e6e6fa',  # Light purple
    3: '#9370db'   # Dark purple
}

# Create a colormap aligned with your unique labels
unique_labels = sorted(df_january['pred_labels'].unique())
custom_cmap = ListedColormap([label_color_map[label] for label in unique_labels])

# Plot with categorical handling enabled
fig, ax = plt.subplots(figsize=(10, 8))
df_january.plot(
    column='pred_labels',
    cmap=custom_cmap,
    categorical=True,  # Critical: tells GeoPandas this is discrete data
    legend=True,
    ax=ax
)

# Optional: Tweak legend labels for clarity
handles, current_labels = ax.get_legend_handles_labels()
ax.legend(handles, [f"Label {label}" for label in unique_labels], title="Predicted Labels")
plt.show()

The categorical=True flag is key here — without it, GeoPandas will try to treat your labels as continuous values, leading to incorrect color mapping and distorted visuals.

Third: Custom Coloring + Legend for color (Continuous Values) #

Your color column is a continuous 0-1 range, so we'll use a gradient colormap and optimize the legend for readability:

import matplotlib.pyplot as plt
from matplotlib.colors import LinearSegmentedColormap

# Create a custom continuous purple gradient
custom_purple_cmap = LinearSegmentedColormap.from_list(
    'my_purples',
    ['#f0e6ff', '#9370db'],  # Light to dark purple
    N=256
)

# Plot with legend customization
fig, ax = plt.subplots(figsize=(10, 8))
color_plot = df_january.plot(
    column='color',
    cmap=custom_purple_cmap,
    legend=True,
    legend_kwds={
        'label': 'Color Intensity',
        'shrink': 0.8,  # Shrink legend to fit better
        'orientation': 'horizontal'  # Optional: horizontal legend for wider plots
    },
    ax=ax
)
plt.show()

Quick Extra Checks #

• If you still see overlaps, check for duplicate geometries with df_january.drop_duplicates(subset='geometry')
• Verify no hexagons are overlapping with df_january.geometry.overlay(df_january, how='intersection').empty — if this returns False, you have overlapping shapes that need cleaning up

内容的提问来源于stack exchange，提问作者John Stud