如何在 Seaborn 的误差带中绘制所有线？

Question

I use this to create a line plot with error bands similar to this: https://seaborn.pydata.org/examples/errorband_lineplots.html

g = sns.lineplot(data=df, x="ann_num_parts", y="probability", hue="full_ann_num_parts", palette=sns.color_palette("hls", 6), err_style="band")

However, I want to plot the actual lines in the error band, ie every row in the data, with a lighter color (lower alpha) than the line of the mean. How do I do this? I'm pretty lost, so any high-level idea or relevant documentation would help. Thanks!

Answer 1

Since there is no data presented, I have taken a sample from the official reference and applied your code. The customization of the error bands cannot be achieved using only Seaborn, so I use matplotlib to make the changes. Set any color for polly collection and line2D. At the same time, set the alpha value to 1.0 for the polly collection, since the alpha value is set low as per default. In addition, the legend handler is still in its original color, so create a line of the specified color and set it to the legend. The manipulation of objects in matplotlib could be written more smartly.

import matplotlib.pyplot as plt
from matplotlib.collections import PolyCollection
from matplotlib.lines import Line2D
import seaborn as sns
sns.set_theme(style="darkgrid")

custom_lines = [Line2D([0], [0], color='#ffc0cb', lw=2), Line2D([0], [0], color='#00ffff', lw=2)]

# Load an example dataset with long-form data
fmri = sns.load_dataset("fmri")

# Plot the responses for different events and regions
g = sns.lineplot(x="timepoint", y="signal",
                 hue="event", data=fmri,
                 palette=sns.color_palette("hls", 2), err_style="band")
#print(g.get_children())

colors = ['#ff0000','#ff0000','#1e90ff','#1e90ff']
colors2 = ['#ffc0cb','#00ffff','#ffc0cb','#00ffff']
i = 0
for p in g.get_children():
    if type(p).__name__ == 'PolyCollection':
        #print(p.get_facecolor())
        p.set_color(colors[i])
        p.set_alpha(1.0)
    if type(p).__name__ == 'Line2D':
        #print(p.get_color())
        p.set_color(colors2[i])
        i += 1
_, labels = g.get_legend_handles_labels()
g.legend(custom_lines, labels)
plt.show()