Matplotlib雷达图标签旋转问题：如何让标签指向圆心？

我尝试用Matplotlib绘制雷达图，希望每个标签从对应角度指向圆心，但当前代码实现的标签旋转效果不符合预期。以下是我的代码：

import math
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

df = pd.read_csv('./results/zh_en_annotation_eval.csv', sep=',',
                 index_col=0, header='infer', encoding='utf-8')

labels = df.columns.values.tolist()
N = len(labels)
values = df.loc['r'].values.tolist()

angles = [n / float(N) * 2 * math.pi for n in range(N)] + [0]


def plot_radar_chart(ax, angles, values, label, color, alpha=0.2):
    values += values[:1]
    ax.plot(angles, values, linewidth=1, linestyle='solid', label=label, color=color)
    ax.fill(angles, values, alpha=alpha, color=color)


fig, ax = plt.subplots(figsize=(16, 12), subplot_kw={'projection': 'polar'})

ax.set_theta_offset(math.pi / 2)
ax.set_theta_direction(-1)
ax.set_xticks(angles[:-1])

for i, angle in enumerate(angles[:-1]):
    angle_deg = np.degrees(angle)
    rotation = angle_deg + 90
    label_position = 1.1
    ax.text(angle, label_position, labels[i], transform=ax.transData,
            ha='center', va='center', rotation=rotation, color='grey', size=10)

maximum = math.ceil(df.loc['r'].max())
minimum = 0.0
ticks = np.linspace(minimum, maximum, 3)
plt.yticks(ticks, labels=[str(tick) for tick in ticks], color='grey', size=7)
plt.ylim(minimum, maximum)

plot_radar_chart(ax, angles, values, label='r', color='red')

ax.set_xticklabels([])
plt.title(f"En-Zh Pearson Correlation", size=11, y=1.1)

plt.legend(loc='upper right', bbox_to_anchor=(1.1, 1.1))
plt.savefig('./results/zh_en_annotation_rader_plot.png')

问题根源 #

原代码的旋转角度计算未考虑极坐标的偏移（theta_offset）和方向（theta_direction）设置，导致标签旋转方向错误，无法指向圆心。

修正后的代码 #

修改标签绘制的循环部分即可解决问题，完整代码如下：

import math
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

df = pd.read_csv('./results/zh_en_annotation_eval.csv', sep=',',
                 index_col=0, header='infer', encoding='utf-8')

labels = df.columns.values.tolist()
N = len(labels)
values = df.loc['r'].values.tolist()

angles = [n / float(N) * 2 * math.pi for n in range(N)] + [0]


def plot_radar_chart(ax, angles, values, label, color, alpha=0.2):
    values += values[:1]
    ax.plot(angles, values, linewidth=1, linestyle='solid', label=label, color=color)
    ax.fill(angles, values, alpha=alpha, color=color)


fig, ax = plt.subplots(figsize=(16, 12), subplot_kw={'projection': 'polar'})

ax.set_theta_offset(math.pi / 2)
ax.set_theta_direction(-1)
ax.set_xticks(angles[:-1])

# 修正标签旋转逻辑
for i, angle in enumerate(angles[:-1]):
    rot_angle = np.degrees(angle)
    # 左半部分标签翻转180度，避免文本倒置
    if angle > np.pi:
        rot_angle += 180
    # 调整角度使标签垂直于径向线并指向圆心
    rot_angle -= 90
    label_position = 1.1
    ax.text(angle, label_position, labels[i],
            ha='center', va='center', 
            rotation=rot_angle, color='grey', size=10)

maximum = math.ceil(df.loc['r'].max())
minimum = 0.0
ticks = np.linspace(minimum, maximum, 3)
plt.yticks(ticks, labels=[str(tick) for tick in ticks], color='grey', size=7)
plt.ylim(minimum, maximum)

plot_radar_chart(ax, angles, values, label='r', color='red')

ax.set_xticklabels([])
plt.title(f"En-Zh Pearson Correlation", size=11, y=1.1)

plt.legend(loc='upper right', bbox_to_anchor=(1.1, 1.1))
plt.savefig('./results/zh_en_annotation_rader_plot.png')

关键修改点 #

1. 旋转角度计算：
   • 将弧度转换为度数后，对左半部分（角度大于np.pi）的标签额外增加180度，避免文本倒置
   • 减去90度，让文本垂直于径向线，确保标签朝向圆心
2. 移除不必要的transform：删除transform=ax.transData，极坐标下直接使用rotation参数即可正确应用旋转

内容的提问来源于stack exchange，提问作者Bo Peng