Creating and Annotating a Grouped Barplot in Python

Since version 3.4. of Matplotlib, a new method Axes.bar_label has been added to support this kind of labeling task. This method takes as parameter a BarContainer containing the bars and usually obtained as an output of plt.bar. To plot a grouped bar chart, bar function of pandas is more convenient ; but the latter does not return a BarContainer. Here is then a painless solution:

import pandas as pd
import matplotlib.pyplot as plt # 3.4.

df = ...
df.plot.bar()

ax = plt.gca()
for container in ax.containers:
    ax.bar_label(container, padding=3)

Let's customize it a bit:

fig, ax = plt.subplots(figsize=(8, 4))

df.plot.bar(rot=0, ax=ax, zorder=2,
    color=["cornflowerblue", "yellowgreen", "gold", "salmon"])
for container in ax.containers:
    ax.bar_label(container, padding=3)

plt.legend(bbox_to_anchor=(0., 1.02, 1., .102), loc='lower left',
     ncol=4, mode="expand", borderaxespad=0.)
plt.grid(True, axis='y', c="lightgrey", zorder=0)
ax.spines['top'].set_color('none')
ax.spines['right'].set_color('none')

There are other ways to convert the data format to a vertical format, but we will draw a bar chart for that vertical data. Then get the x-axis position and height of that bar, and annotate it. In my code, I have placed the text at half the height.

df_long = df.unstack().to_frame(name='value')
df_long = df_long.swaplevel()
df_long.reset_index(inplace=True)
df_long.columns = ['group', 'status', 'value']

import matplotlib.pyplot as plt

fig, ax = plt.subplots(figsize=(12, 8))

g = sns.barplot(data=df_long, x='group', y='value', hue='status', ax=ax)

for bar in g.patches:
    height = bar.get_height()
    ax.text(bar.get_x() + bar.get_width() / 2., 0.5 * height, int(height),
                ha='center', va='center', color='white')

plt.show()

It's important to iterate through each of the columns and plot them as you go in a for-loop. Then reassign the column names. Here is a full example. You should be able to copy and paste this code, then run it.

import numpy as np
import pandas as pd
from matplotlib import pyplot as plt

# your data
df = pd.DataFrame(
    {
        'Early': {'A': 824, 'B': 701, 'C': 1050, 'D': 764, 'E': 993},
        'Long Overdue': {'A': 238, 'B': 270, 'C': 489, 'D': 549, 'E': 471},
        'On Time': {'A': 1021, 'B': 1025, 'C': 120, 'D': 71, 'E': 57},
        'Overdue': {'A': 493, 'B': 580, 'C': 917, 'D': 1192, 'E': 1055}
    })

# set the figure size
fig, ax = plt.subplots(figsize = (10,7), dpi = 200)

# select the colors you would like to use for each category
colors = ['skyblue','goldenrod','slateblue','seagreen']

# used to set the title, y, and x labels
ax.set_title('\nTrain Times\n', fontsize = 14)
ax.set_xlabel('\nCategories\n', fontsize = 14)
ax.set_ylabel('\nValues\n', fontsize = 14)

# create an offsetting x axis to iterate over within each group
x_axis = np.arange(len(df))+1 

# center each group of columns
offset = -0.3                 

# iterate through each set of values and the colors associated with each 
# category
for index, col_name, color in zip(x_axis, df.columns, colors):
    
    x = x_axis+offset
    height = df[col_name].values

    ax.bar(
        x, 
        height, 
        width = 0.2, 
        color = color, 
        alpha = 0.8, 
        label = col_name
    )   
    
    offset += 0.2
    
    # set the annotations
    props = dict(boxstyle='round', facecolor='white', alpha=1)
    
    for horizontal, vertical in zip(x, height):
        
        ax.text(
            horizontal-len(str(vertical))/30, 
            vertical+26, 
            str(vertical), 
            fontsize=12, 
            bbox=props)
        

# set the y limits so the legend appears above the bars
ax.set_ylim(0, df.to_numpy().max()*1.25)

# relabel the x axis
ax.set_xticks(x_axis)                   # offset values
ax.set_xticklabels(df.index.to_list())  # set the labels for each group

# the legend can be set to multiple values. 'Best' has Matplotlib automatically set the location.
# setting ncol to the length of the dataframe columns sets the legend horizontally by the length 
# of the columns
plt.legend(loc = 'best', ncol=len(df.columns), fontsize = 12)                        
plt.show()

This should give you the following plot. I tried to match the colors as closely as possible to the picture you provided. However, you can choose your own. Here is a link to a chart and list of all the colors available for you to use. Chart of available Matplotlib colors