Plotly（Python）自定义色阶重置为默认值的问题及需求

Hey there! I’ve dealt with this exact challenge before—getting a RdBu color scale anchored at z=1 (your mean value) while preserving full dynamic range and keeping sharp color detail in population-dense regions. Here’s a step-by-step solution that works:

Core Approach #

The main trick is to build a custom color scale that locks yellow (the RdBu midpoint) exactly to z=1, while adding extra color stops in data-dense areas to avoid washing out subtle differences. We’ll calculate normalized positions for each color stop based on your actual z-value min/max, so the scale adapts to your dataset.

Full Code Example #

Let’s assume you have your state data ready, with z_values being each state’s value divided by the overall mean (so z=1 is the average):

import plotly.express as px
import numpy as np

# Replace with your actual state codes and z-values
state_codes = ['AL', 'AK', 'AZ', 'AR', 'CA', 'CO', 'CT', 'DE', 'FL', 'GA',
               'HI', 'ID', 'IL', 'IN', 'IA', 'KS', 'KY', 'LA', 'ME', 'MD',
               'MA', 'MI', 'MN', 'MS', 'MO', 'MT', 'NE', 'NV', 'NH', 'NJ',
               'NM', 'NY', 'NC', 'ND', 'OH', 'OK', 'OR', 'PA', 'RI', 'SC',
               'SD', 'TN', 'TX', 'UT', 'VT', 'VA', 'WA', 'WV', 'WI', 'WY']
z_values = np.random.uniform(0.4, 2.8, size=len(state_codes))  # Example z-scores relative to mean=1

# Calculate min/max of your z-values to normalize color stops
z_min = z_values.min()
z_max = z_values.max()

# Build custom color scale: anchor yellow at z=1, add stops for dense regions
# Format: [normalized_position, hex_color]
color_stops = [
    # Lower range (z < 1): blue → yellow
    [(z_min - z_min)/(z_max - z_min), '#003399'],  # Dark blue for lowest z
    [(0.7 - z_min)/(z_max - z_min), '#6699FF'],
    [(0.9 - z_min)/(z_max - z_min), '#CCDDFF'],
    [(1 - z_min)/(z_max - z_min), '#FFFF00'],  # Exact midpoint at z=1 (yellow)
    # Upper range (z > 1): yellow → red
    [(1.1 - z_min)/(z_max - z_min), '#FFDDCC'],
    [(1.3 - z_min)/(z_max - z_min), '#FF9966'],
    [(1.6 - z_min)/(z_max - z_min), '#FF6633'],
    [(z_max - z_min)/(z_max - z_min), '#990000']   # Dark red for highest z
]

# Create the choropleth
fig = px.choropleth(
    locations=state_codes,
    locationmode='USA-states',
    color=z_values,
    scope='usa',
    color_continuous_scale=color_stops,
    color_continuous_midpoint=None,  # Disable auto midpoint to use our custom scale
    range_color=[z_min, z_max]  # Lock to full z-value range (no clipping)
)

# Add clear colorbar labeling to highlight the mean
fig.update_layout(
    coloraxis_colorbar=dict(
        title='Value / Mean',
        tickvals=[z_min, 1, z_max],
        ticktext=[f'{z_min:.2f}', '1 (Mean)', f'{z_max:.2f}'],
        ticks='outside',
        thickness=20
    )
)

fig.show()

Key Details Explained #

1. Custom Color Stops: Each stop uses a normalized position calculated as (z_value - z_min)/(z_max - z_min). This ensures z=1 maps to a fixed position on the color bar, regardless of how asymmetric your min/max values are.
2. Extra Stops for Dense Areas: By adding stops at z=0.7, 0.9, 1.1, etc., we increase color resolution in regions where population-dense states are likely to cluster (near the mean or high z-values). This makes small differences easier to spot without losing full dynamic range.
3. Locking the Range: range_color=[z_min, z_max] guarantees the scale covers every value in your dataset—no clipping at either end.
4. Disabling Auto Midpoint: Setting color_continuous_midpoint=None prevents Plotly from overriding your custom scale with its default midpoint calculation.

Adjustments for Your Data #

If your dense regions are concentrated in high z-values (e.g., populous states have z > 1), shift extra stops to the upper range:

# Modified stops for high-value dense regions
color_stops = [
    [(z_min - z_min)/(z_max - z_min), '#003399'],
    [(1 - z_min)/(z_max - z_min), '#FFFF00'],
    [(1.2 - z_min)/(z_max - z_min), '#FFDDCC'],
    [(1.4 - z_min)/(z_max - z_min), '#FFBB99'],
    [(1.7 - z_min)/(z_max - z_min), '#FF8866'],
    [(2.0 - z_min)/(z_max - z_min), '#FF5533'],
    [(z_max - z_min)/(z_max - z_min), '#990000']
]

内容的提问来源于stack exchange，提问作者NauticalN