如何基于嵌套列表数据绘制堆叠条形图？

Got it, let's build this stacked bar chart that adapts to your dynamic cluster data. We'll use matplotlib since it gives us full control over stacking segments and mapping consistent colors to clusters, even when the number of clusters changes.

Step 1: Prepare the Data & Import Libraries #

First, let's start by importing the tools we need and defining your input lists:

import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
import numpy as np

# Your input data
totalpointperxaxis = [6, 9, 13, 5, 14, 382, 26, 2, 45, 2]
clusternamesList = [['Cluster1', 'Cluster2'], ['Cluster1', 'Cluster3'], ['Cluster2', 'Cluster4'], ['Cluster1', 'Cluster3'], ['Cluster2', 'Cluster5'], ['Cluster3', 'Cluster6', 'Cluster7'], ['Cluster2', 'Cluster4', 'Cluster6', 'Cluster7'], ['Cluster1', 'Cluster3'], ['Cluster1', 'Cluster2', 'Cluster4', 'Cluster5', 'Cluster6'], ['Cluster1', 'Cluster3']]
ppclusterList = [[1, 5], [4, 5], [12, 1], [1, 4], [13, 1], [6, 173, 203], [21, 2, 1, 2], [1, 1], [2, 34, 2, 6, 1], [1, 1]]

Step 2: Collect Unique Clusters & Assign Colors #

We need to gather all unique cluster names first, then map each to a distinct color. We'll use a colormap that has enough colors for potential future clusters (like tab20 which has 20 distinct hues):

# Collect all unique cluster names
all_clusters = []
for clusters in clusternamesList:
    all_clusters.extend(clusters)
unique_clusters = list(set(all_clusters))
unique_clusters.sort()  # Sort for consistent ordering

# Create a color map for clusters
num_clusters = len(unique_clusters)
cmap = plt.get_cmap('tab20', num_clusters)
cluster_color_map = {cluster: cmap(i) for i, cluster in enumerate(unique_clusters)}

Step 3: Plot the Stacked Bars #

We'll iterate over each bar (x-axis position), stack each cluster's segment on top of the previous one. We'll track the bottom position for each bar to build the stack:

# Set up the plot
fig, ax = plt.subplots(figsize=(12, 6))

# X-axis positions (one per bar)
x_positions = np.arange(len(totalpointperxaxis))

# Initialize bottom for stacking
bottom = np.zeros(len(totalpointperxaxis))

# Iterate through each cluster to plot its segments across all bars
for cluster in unique_clusters:
    # Create an array of values for this cluster (0 if the cluster isn't present in a bar)
    cluster_values = []
    for names, points in zip(clusternamesList, ppclusterList):
        if cluster in names:
            idx = names.index(cluster)
            cluster_values.append(points[idx])
        else:
            cluster_values.append(0)
    
    # Plot the segment for this cluster
    ax.bar(x_positions, cluster_values, bottom=bottom, label=cluster, color=cluster_color_map[cluster])
    
    # Update bottom for the next layer
    bottom += np.array(cluster_values)

Step 4: Customize the Plot #

Add labels, titles, adjust ticks, and handle the legend (since clusters might be many, we'll place it outside the plot):

# Customize axes and labels
ax.set_xlabel('Bar Index')
ax.set_ylabel('Total Points')
ax.set_title('Stacked Bar Chart: Points per Cluster')
ax.set_xticks(x_positions)
ax.set_xticklabels([f'Bar {i+1}' for i in x_positions])

# Add legend outside the plot to avoid overlapping
ax.legend(title='Clusters', bbox_to_anchor=(1.05, 1), loc='upper left')

# Adjust layout to fit the legend
plt.tight_layout()

# Show the plot
plt.show()

Key Notes for Dynamic Data #

• Automatic Adaptation: If you add more bars to your lists, the code will automatically create new x positions—no hardcoding needed.
• New Clusters: Any new clusters added to clusternamesList will be automatically picked up by the all_clusters collection, assigned a new color, and included in the plot/legend.
• Data Consistency Check: For peace of mind, you can add a quick check to ensure each bar's total points match totalpointperxaxis:

  # Verify data consistency (optional)
  for i, (total, points) in enumerate(zip(totalpointperxaxis, ppclusterList)):
      if sum(points) != total:
          print(f"Warning: Bar {i+1} total points ({total}) doesn't match sum of cluster points ({sum(points)})")
  

This code will generate a stacked bar chart exactly like you described, with each cluster's segment colored consistently across all bars, and it will adapt seamlessly as your data grows or changes.

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