为Seaborn热图/聚类图添加第二个颜色条的技术求助

Hey there! I feel your pain—spending hours troubleshooting a small visualization detail is never fun. Let's fix this: the issue here is that seaborn.clustermap doesn't automatically generate a colorbar for the row_colors parameter, since it's primarily meant for grouping/annotating rows. We need to manually create a color mapping and attach it to the figure.

Here's how to do it, depending on whether your vertical bar represents continuous values or categorical labels:

Case 1: Continuous Values in the Vertical Bar #

If your row_colors are based on a continuous dataset (like your label array being numerical), use this approach:

import pickle
import numpy as np
import seaborn as sns
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.cm import ScalarMappable
from matplotlib.colors import Normalize

# Load your data
feat_mat, freq, label = pickle.load(open('file.pkl', 'rb'))
feat_mat_df = pd.DataFrame(feat_mat[4])

# Define your row colors (assuming label is the continuous data for the blue bar)
row_colors = label

# Draw the clustermap
g = sns.clustermap(feat_mat_df, row_colors=row_colors, cmap='Blues')  # Use your desired cmap

# Create a ScalarMappable to link values to the color map
norm = Normalize(vmin=row_colors.min(), vmax=row_colors.max())
sm = ScalarMappable(norm=norm, cmap='Blues')
sm.set_array([])  # We just need the color mapping, not actual data

# Add the colorbar aligned with the vertical row bar
cbar = g.fig.colorbar(sm, ax=g.ax_row_dendrogram, orientation='vertical', pad=0.01)
# Customize the colorbar label
cbar.set_label('Your Value Description')

plt.show()

Case 2: Categorical Labels in the Vertical Bar #

If your vertical bar is for categorical groups (e.g., different classes in label), adjust the code to map categories to colors:

import pickle
import numpy as np
import seaborn as sns
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.cm import ScalarMappable
from matplotlib.colors import Normalize, ListedColormap

# Load your data
feat_mat, freq, label = pickle.load(open('file.pkl', 'rb'))
feat_mat_df = pd.DataFrame(feat_mat[4])

# Map categorical labels to numerical values
unique_labels = np.unique(label)
label_to_num = {lab: idx for idx, lab in enumerate(unique_labels)}
num_labels = [label_to_num[lab] for lab in label]

# Create a custom colormap for your categories (using Blues palette here)
cmap = ListedColormap(sns.color_palette('Blues', n_colors=len(unique_labels)))

# Draw the clustermap
g = sns.clustermap(feat_mat_df, row_colors=num_labels, cmap=cmap)

# Set up the color mapping
norm = Normalize(vmin=0, vmax=len(unique_labels)-1)
sm = ScalarMappable(norm=norm, cmap=cmap)
sm.set_array([])

# Add the colorbar with category labels
cbar = g.fig.colorbar(sm, ax=g.ax_row_dendrogram, orientation='vertical', pad=0.01)
cbar.set_ticks(np.arange(len(unique_labels)))
cbar.set_ticklabels(unique_labels)
cbar.set_label('Category Name')

plt.show()

Why Your Previous Attempts Didn't Work #

Calling plt.colorbar(row_colors) directly fails because row_colors is just an array of values—it doesn't carry any information about the color map or value range. By creating a ScalarMappable, we explicitly link your values to the color palette, which lets matplotlib generate a proper colorbar that aligns with your clustermap.

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