多类别Plotly箱线图优化：默认颜色获取、单图例设置及箱线顺序控制

Hey there! Let's tackle your three Plotly box plot customization questions one by one, with a polished code solution that addresses all your needs:

1. Use Plotly's Default Color Cycle #

Instead of hardcoding colors, you can access Plotly's built-in qualitative color palette via plotly.colors.qualitative.Plotly — this is the exact palette used by default for traces, so your charts will match other Plotly visualizations seamlessly. Just pick the first n colors you need (in this case, 2 for categories "a" and "b"). You can also adjust transparency for fill colors by modifying the color string to use rgba instead of rgb.

2. More Flexible Legend Control #

Instead of checking if you're on the first value column, use legend groups to link traces of the same category across subplots. Then, only enable the legend for the first occurrence of each category. This approach is far more robust if you ever add more subplots or reorder your value columns, and it also ensures clicking a legend item toggles all traces of that category across both subplots.

3. Control Category Display Order #

To specify the exact order of categories ("a" vs "b") in your box plots, define a custom order list and set categoryorder='array' + categoryarray=your_order in each Box trace. This forces Plotly to render categories in your desired sequence instead of relying on default alphabetical ordering.

Full Modified Code #

import pandas as pd
import numpy as np
from plotly import graph_objects as go
from plotly.subplots import make_subplots
import plotly.colors  # Import to access default color palette

np.random.seed(42)
df = pd.DataFrame(
    {
        "val1": np.random.normal(0, 1, size=100),
        "val2": np.random.normal(5, 2, size=100),
        "cat": np.random.choice(["a", "b"], size=100),
    }
)

# --- Customization Settings ---
# 1. Grab first 2 colors from Plotly's default qualitative palette
default_colors = plotly.colors.qualitative.Plotly
# Optional: Add transparency to fill colors
fill_colors = {
    "a": default_colors[0].replace("rgb", "rgba").replace(")", ", 0.5)"),
    "b": default_colors[1].replace("rgb", "rgba").replace(")", ", 0.5)")
}
# Use this if you don't want transparency:
# fill_colors = {"a": default_colors[0], "b": default_colors[1]}

# 2. Define your preferred category display order
cat_order = ["a", "b"]  # Swap to ["b", "a"] to reverse the order

# 3. Track which categories already have a legend entry
legend_added = set()

# Create subplots
fig = make_subplots(rows=2, cols=1, subplot_titles=["Value 1 dist", "Value 2 dist"])

for i, val in enumerate(["val1", "val2"]):
    # Iterate through categories in your custom order
    for c in cat_order:
        dff = df[df["cat"] == c]
        # Show legend only if this category hasn't been added yet
        show_legend = c not in legend_added
        
        fig.add_trace(
            go.Box(
                y=dff[val],
                x=dff["cat"],
                boxmean="sd",
                name=c,
                showlegend=show_legend,
                legendgroup=c,  # Link same-category traces across subplots
                fillcolor=fill_colors[c],
                line={"color": fill_colors[c].split(", 0.5)")[0] + ")"},  # Opaque border
                # Enforce custom category order on the x-axis
                categoryorder='array',
                categoryarray=cat_order
            ),
            row=i + 1,
            col=1,
        )
        
        # Mark category as having its legend entry added
        if show_legend:
            legend_added.add(c)

# Optional: Tweak layout for better readability
fig.update_layout(
    height=600,
    title_text="Customized Box Plots",
    legend_title_text="Categories"
)

fig.show()

Key Details #

• Default Colors: The plotly.colors.qualitative.Plotly list includes 10 distinct colors, so you can easily expand this for more categories.
• Legend Groups: This feature creates a unified legend experience — clicking "a" hides both "a" boxes in the top and bottom subplots.
• Category Order: Simply modify the cat_order list to reposition categories instantly, no need to rearrange your data.

内容的提问来源于stack exchange，提问作者Dror