How to add labels to two overlaid bar plots showing value_counts in pandas

Question

I have two overlaid bar plots from binary column data. I need to add the value_counts number as label to each bar of the plot.

df = pd.DataFrame({'y_pred': y_pred, 'y_test': y_test})
df.y_test.value_counts().plot(kind='bar')
df.y_pred.value_counts().plot(kind='bar', width=0.3, color='r')
plt.legend()
plt.show()

In another question I found:

for p in ax.patches:
        ax.annotate(str(p.get_height()), (p.get_x() * 1.005, p.get_height() * 1.005))

but I did not understand how this would loop through the bars in my case. Copy paste obviously did not work. the code with 'for loop' was executed without error messages but still showed the same plot without labels...

df = pd.DataFrame({'y_pred': y_pred, 'y_test': y_test})
df.y_test.value_counts().plot(kind='bar')
df.y_pred.value_counts().plot(kind='bar', width=0.3, color='r')
for p in ax.patches:
    ax.annotate(str(p.get_height()), (p.get_x() * 1.005, p.get_height() * 1.005))

plt.legend()

plt.show()

Answer 1

You need to pass ax into the plot command:

fig, ax = plt.subplots()
df.y_test.value_counts().plot(kind='bar', ax=ax)
df.y_pred.value_counts().plot(kind='bar', width=0.3, color='r', ax=ax)
for p in ax.patches:
    ax.annotate(str(p.get_height()), (p.get_x() * 1.005, p.get_height() * 1.005))

plt.legend()

plt.show()

Output:

Play with get_width a little to align the labels.