Plotting ax.bar_label on sns does not iterate through hue appropriately-CodePudding

I have this table

print(tmp.head())

  tumor_type tumor_or_normal           call  size    cancer_n     %
0        ACC           Tumor  Amplification     2    ACC\nn=5  40.0
1        ACC           Tumor       Deletion     1    ACC\nn=5  20.0
2        ACC           Tumor        Diploid     2    ACC\nn=5  40.0
3       BLCA           Tumor  Amplification     3  BLCA\nn=10  30.0
4       BLCA           Tumor       Deletion     1  BLCA\nn=10  10.0

for which I am trying to add the size column values as bar labels:

plt.figure(figsize = (20,8))
palette = {'Amplification': 'red', 'Deletion': 'blue', 'Diploid': 'grey'}
ax = sns.barplot(x ="cancer_n", y = '%', data = tmp, hue = "call", palette=palette, alpha=0.7)
for container in ax.containers:
    ax.bar_label(container, labels=tmp['size'])
plt.show()

This uses the correct numbers but labels the bars incorrectly (labelling the first bar of each x-axis label instead of each bar of hue variable)

See here for the plot and error

How can I correct this?

CodePudding user response：

The issue is that you give the whole size column for each hue. You need to provide the size values for 'Amplification', 'Deletion', and 'Diploid' separately.

plt.figure(figsize = (20,8))
palette = {'Amplification': 'red', 'Deletion': 'blue', 'Diploid': 'grey'}
ax = sns.barplot(x ="cancer_n", y = '%', data = tmp, hue = "call", palette=palette, alpha=0.7)
for container in ax.containers:
    tmp_hue = tmp.loc[tmp['call']==container.get_label()]
    ax.bar_label(container, labels=tmp_hue['size'])
plt.show()