Python如何绘制含winner_rank_status及获胜方颜色占比的堆叠柱状图

你当前的代码是对每个winner_rank_status分组单独做了占比归一化，导致所有柱子总高度均为100%，按照需求修改后的代码如下：

import pandas as pd
import matplotlib.pyplot as plt

# 1. 统计每个(winner_rank_status, winner)组合的样本数
Q3_count = games_df.groupby(['winner_rank_status', 'winner']).size().reset_index(name='count')
# 2. 计算每个组合占总样本的比例，堆叠后总高度即为对应winner_rank_status的整体占比
total_sample = Q3_count['count'].sum()
Q3_count['total_pct'] = round(100 * Q3_count['count'] / total_sample, 2)
# 3. 计算每个获胜方在所属rank分组内的占比，用于柱子内部标注
Q3_count['inner_pct'] = Q3_count.groupby('winner_rank_status')['count'].apply(lambda x: round(100 * x / x.sum(), 2))
# 4. 转换为绘图需要的宽表格式
Q3_df = Q3_count.pivot(index='winner_rank_status', columns='winner', values='total_pct').fillna(0)

# 绘制堆叠柱状图
ax = Q3_df.plot(
    kind='bar',
    stacked=True,
    figsize=(14,7),
    rot=0,
    title='Effect of piece colour and winner rating status on the result',
    color=['black','grey','white'],
    edgecolor='black'
)

# 按顺序生成内部标注内容
inner_labels = Q3_count.sort_values(['winner_rank_status', 'winner'])['inner_pct'].values
label_index = 0
for container in ax.containers:
    labels = [f"{val}%" for val in inner_labels[label_index:label_index+len(container)]]
    ax.bar_label(container, labels=labels, label_type='center', color='b')
    label_index += len(container)

# 可选：在柱子顶部添加当前rank分类的整体占比标注
total_pct_per_group = Q3_df.sum(axis=1).values
ax.bar_label(ax.containers[-1], labels=[f"{round(val,2)}%" for val in total_pct_per_group], label_type='edge', color='r', fontweight='bold')

plt.ylabel('占总样本比例')
plt.show()

效果说明 #

• 每个柱子的总高度为对应winner_rank_status分类占全部样本的真实占比
• 柱子内部蓝色标注为对应颜色获胜方在当前分类下的占比
• 可选添加的顶部红色标注为当前分类的整体占比，方便直观读取数值

内容的提问来源于stack exchange，提问作者Vahid