seaborn visualize point count in point plot

Question

How can I add a point count in the point plot in seaborn?

df = pd.DataFrame({'group':[0,1,1],'percentage':[0.5, .3,.7],'value_group':[1,2,3], 'count':[1, 10, 20]})
df['value_group'] = pd.qcut(df.value_group, 2)
   group  percentage  value_group  count
0      0         0.5            1      1
1      1         0.3            2     10
2      1         0.7            3     20

With

sns.pointplot(x="value_group", y="percentage", hue="group", data=df)

I get: But instead I would want:

How can this be achieved in seaborn?

edit

They are similar, but not the same. My value_group is obtained using pd.qcut and the code referenced i.e. cannot handle these.

python Seaborn - annotations with pointplot

import matplotlib.pyplot as plt
import seaborn as sns

tips = sns.load_dataset("tips")
ax = sns.pointplot(x="time", y="total_bill", hue="smoker", data=tips)

for c in ax.collections:
    for of in c.get_offsets():
        ax.annotate("Label", of)

plt.show()

looks interesting, but so far I do not yet know how to match the right labels/indices with my counts.

Possible duplicate of [How to add counts of points as a label in a sparse scatter plot](https://stackoverflow.com/questions/43539011/how-to-add-counts-of-points-as-a-label-in-a-sparse-scatter-plot) — Vishnu Dasu, Jun 10 '19 at 06:58

Sheldore · Accepted Answer · 2019-06-10T15:11:45.583

The problem was bit nontrivial that I was thinking. Since you have used hue="group", there are two groups in the plot and therefore, ax.collections has a length of 2. So to have the annotations in the correct order, I used a tacking index j.

You can zip the offsets and the DataFrame values you want to show, and annotate them using a for loop as

import matplotlib.pyplot as plt
import seaborn as sns
sns.set()

df = pd.DataFrame({'group':[0,1,1],'percentage':[0.5, .3,.7],'value_group':[1,2,3], 'count':[1, 10, 20]})

ax = sns.pointplot(x="value_group", y="percentage", hue="group", data=df)

j = 0 # <--- Index to keep rack of values
values = df['count'].values # <--- store the values in a variable for easy access

for c in ax.collections:
    for i, of in zip(range(len(c.get_offsets())), c.get_offsets()):
        ax.annotate(values[j], of, color='red', fontsize=24)
        j += 1
ax.legend(loc=(0.8, 0.1))
plt.show()

seaborn visualize point count in point plot

edit

1 Answers1