使用matplotlib.pyplot绘制分组彩色散点图报错及颜色统一问题求助

Hey there, let's break down why you're seeing that ValueError and get your scatter plot showing distinct colors for your two groups!

The Root Cause #

The error pops up because your y_train contains categorical group labels (like string names) instead of numerical values. Matplotlib's c parameter needs numbers to map to the colormap you chose (plt.cm.Paired)—it can't directly interpret raw category names like "sample group".

Solutions to Fix Color Grouping #

1. Convert Categorical Labels to Numerical Values (Recommended)

Use LabelEncoder from scikit-learn to turn your group labels into numbers that matplotlib can work with:

from sklearn.preprocessing import LabelEncoder
import matplotlib.pyplot as plt

# Transform categorical labels to numerical codes
label_encoder = LabelEncoder()
y_train_numeric = label_encoder.fit_transform(y_train)

# Create your plot with the numeric labels
fig = plt.figure(1, figsize=(10, 6))
plt.scatter(X_train_reduced[:, 0], X_train_reduced[:, 1], 
            c=y_train_numeric, 
            cmap=plt.cm.Paired, 
            linewidths=10)

# Optional: Add a colorbar to clarify which number maps to which group
plt.colorbar(ticks=label_encoder.transform(label_encoder.classes_), 
             label='Sample Group')
plt.clim(-0.5, len(label_encoder.classes_) - 0.5) # Centers ticks on color groups
plt.show()

2. Manually Map Labels to Specific Colors

If you want full control over which color each group gets (instead of using a colormap), create a label-to-color dictionary:

import matplotlib.pyplot as plt

# Replace 'group1'/'group2' with your actual group names
color_mapping = {'group1': '#ff7f0e', 'group2': '#1f77b4'}
# Generate a color list for each data point
point_colors = [color_mapping[label] for label in y_train]

fig = plt.figure(1, figsize=(10, 6))
plt.scatter(X_train_reduced[:, 0], X_train_reduced[:, 1], 
            c=point_colors, 
            linewidths=10)
plt.show()

3. Quick Fix with Pandas factorize

If you're already using pandas, this is a fast way to convert labels to numbers:

import pandas as pd
import matplotlib.pyplot as plt

# Convert labels to numerical codes
y_train_numeric, _ = pd.factorize(y_train)

fig = plt.figure(1, figsize=(10, 6))
plt.scatter(X_train_reduced[:, 0], X_train_reduced[:, 1], 
            c=y_train_numeric, 
            cmap=plt.cm.Paired, 
            linewidths=10)
plt.show()

Quick Note on Your Code #

You're calling plt.figure() twice (fig = plt.figure(1, ...) followed by plt.figure()). This creates two separate figure windows—you can remove one of these lines to keep your plot focused in a single figure.

内容的提问来源于stack exchange，提问作者user7249622