使用 seaborn 使用特定标记标记抖动

Question

我有四个不同的label_list。如果每个label_list 的第二列和第三列的值相同，我想用一个特定的标记来标记那个点。 例如：

df = pd.DataFrame([['0', 0.3, 'gd','gd'],['0',0.2,'gf','gd'],['1',0.5,'gd','gd'],['1',0.1,'gf','gf'],['2', 0.9,'gd','gd'],['2', 0.1,'gf','gf'],
                   ['0', 0.8,'gd','gd'],['0',0.3,'gf','gf'],['1',0.3,'gd','gd'],['1',0.3,'gf','gf'],['2', 0.4,'gd','gf'],['2', 0.3,'gf','gf'],
                   ['0', 0.4,'gd','gf'],['0',0.4,'gf','gd'],['1',0.0,'gd','gf'],['1',0.9,'gf','gd'],['2', 0.7,'gd','gf'],['2', 0.4,'gf','gd']])

df.columns = ['label', 'score','hotel','hotel_1']

label_list = ['0', '1', '2','3']

fig = plt.figure(figsize=(8, 5))
g=sns.boxplot(x='label', y='score', data=df, hue='hotel',dodge =True)
g=sns.stripplot(x='label',y = 'score',data=df, hue='hotel_1',dodge =False)
g.legend_.remove()

plt.show()

Answer 1

您可以分两步绘制标记，一次在最后一列相等时，一次在它们不同时。由于默认标记可能太小，size 参数会很方便：

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns

df = pd.DataFrame([['0', 0.3, 'gd', 'gd'], ['0', 0.2, 'gf', 'gd'], ['1', 0.5, 'gd', 'gd'], ['1', 0.1, 'gf', 'gf'], ['2', 0.9, 'gd', 'gd'], ['2', 0.1, 'gf', 'gf'],
                   ['0', 0.8, 'gd', 'gd'], ['0', 0.3, 'gf', 'gf'], ['1', 0.3, 'gd', 'gd'], ['1', 0.3, 'gf', 'gf'], ['2', 0.4, 'gd', 'gf'], ['2', 0.3, 'gf', 'gf'],
                   ['0', 0.4, 'gd', 'gf'], ['0', 0.4, 'gf', 'gd'], ['1', 0.0, 'gd', 'gf'], ['1', 0.9, 'gf', 'gd'], ['2', 0.7, 'gd', 'gf'], ['2', 0.4, 'gf', 'gd']])

df.columns = ['label', 'score', 'hotel', 'hotel_1']

label_list = ['0', '1', '2', '3']
fig = plt.figure(figsize=(8, 5))
palette={'gd': 'steelblue', 'gf': 'darkorange'}
g = sns.boxplot(x='label', y='score', data=df, hue='hotel', dodge=True, palette=palette)
g = sns.stripplot(x='label', y='score', data=df[df['hotel'] == df['hotel_1']], hue='hotel_1',
                  dodge=False, marker='*', size=10, palette=palette)
g = sns.stripplot(x='label', y='score', data=df[df['hotel'] != df['hotel_1']], hue='hotel_1',
                  dodge=False, marker='D', size=7, palette=palette)

g.legend_.remove()

plt.show()

PS：为stripplots 设置dodge=True 并更改hue_order：

palette={'gd': 'steelblue', 'gf': 'darkorange'}
g = sns.boxplot(x='label', y='score', data=df, hue='hotel', dodge=True, palette=palette, hue_order=['gd', 'gf'])
g = sns.stripplot(x='label', y='score', data=df[df['hotel'] == df['hotel_1']], hue='hotel_1', hue_order=['gf', 'gd'],
                  dodge=True, marker='*', size=10, palette=palette)
g = sns.stripplot(x='label', y='score', data=df[df['hotel'] != df['hotel_1']], hue='hotel_1', hue_order=['gf', 'gd'],
                  dodge=True, marker='D', size=7, palette=palette)