Seaborn lmplot配置求助：设置y轴显示分段范围

Got it, let's fix this up so your y-axis only shows the extreme correlation ranges (-1.0 to -0.8 and 0.8 to 1.0) by breaking the middle section. Here's how to adjust your existing code to make this happen:

Modified Code #

import seaborn as sns
import matplotlib.pyplot as plt
from matplotlib.gridspec import GridSpec

# Your original lmplot setup (kept all your original parameters)
g = sns.lmplot(
    x='Aantal Sterkst Negatief',  # Horizontal axis
    y='Corr Sterkst Negatief',    # Vertical axis
    data=df_PAR_metingen_half_1,  # Data source
    fit_reg=False,                # Don't fix a regression line
    hue="Sterkst Negatief",       # Color by category
    scatter_kws={
        "marker": "D",            # Marker style
        "s": 100, 
        'alpha': 0.3
    },
    height=12                     # Base figure size
)

# Set up a broken y-axis using GridSpec to split the plot into two stacked axes
fig = g.fig
# Create 2 rows, 1 column; small gap between axes to show the break
gs = GridSpec(2, 1, height_ratios=[1, 1], hspace=0.05)

# Upper axis: displays 0.8 to 1.0 range
ax1 = fig.add_subplot(gs[0])
# Lower axis: displays -1.0 to -0.8 range
ax2 = fig.add_subplot(gs[1])

# Copy all scatter plot elements from the original lmplot axis to both new axes
original_ax = g.axes[0,0]
for line in original_ax.get_lines():
    ax1.add_line(line)
    ax2.add_line(line)
for collection in original_ax.collections:
    ax1.add_collection(collection)
    ax2.add_collection(collection)

# Define y-axis limits for each segment
ax1.set_ylim(0.8, 1.0)
ax2.set_ylim(-1.0, -0.8)

# Hide connecting spines to visually break the axis
ax1.spines['bottom'].set_visible(False)
ax2.spines['top'].set_visible(False)
# Hide x-ticks on upper axis since they'll be duplicated
ax1.tick_params(axis='x', bottom=False, labelbottom=False)

# Add slanted break markers to clearly indicate the broken axis
d = 0.015  # Size of the break lines
kwargs = dict(transform=ax1.transAxes, color='k', clip_on=False)
ax1.plot((-d, +d), (-d, +d), **kwargs)
ax1.plot((1-d, 1+d), (-d, +d), **kwargs)

kwargs.update(transform=ax2.transAxes)
ax2.plot((-d, +d), (1-d, 1+d), **kwargs)
ax2.plot((1-d, 1+d), (1-d, 1+d), **kwargs)

# Add your original plot styling elements
# Horizontal red line at y=0 (it won't appear in either axis since our limits exclude 0)
ax1.axhline(y=0, c='red', linestyle='dashed', zorder=-1)
ax2.axhline(y=0, c='red', linestyle='dashed', zorder=-1)

# Set title and labels to span both axes
fig.suptitle('Correlations by chemical', fontweight='bold', fontsize=12, y=0.95)
ax2.set_xlabel('Count')
fig.text(0.01, 0.5, 'Correlation Coefficient', va='center', rotation='vertical')

# Adjust layout to avoid overlapping elements
plt.tight_layout()
plt.show()

Quick Breakdown of Key Changes: #

• We use GridSpec to stack two axes with a tiny gap, creating the broken axis effect.
• All your original scatter plot data is copied to both axes so points in the extreme ranges show up where they belong.
• We hide the connecting spines and add slanted lines to make the axis break obvious to viewers.
• Title and labels are adjusted to span both axes for a clean, professional look.

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