Python DataFrame - plot 带有分组列（至少两列）的数据框条形图

Question

I've been struggling to recreate this Excel graph in python using matlplotlib:

The data is in a dataframe; I'm trying to automate the process of generating this graph.

I've tried unstacking my dataframe, subplotting, but I haven't managed to create the "Zone" index which is so elegant in Excel. I have successfully managed to plot the graph without this "Zone" index, but that's not really what I want to do.

Here is my code:

data = pd.DataFrame(
    {
        'Factory Zone':
        ["AMERICAS","APAC","APAC","APAC","APAC","APAC","EMEA","EMEA","EMEA","EMEA"],
        'Factory Name':
        ["Chocolate Factory","Crayon Factory","Jobs Ur Us", "Gibberish US","Lil Grey", "Toys R Us","Food Inc.",
        "Pet Shop", "Bonbon Factory","Carrefour"],
        'Production Day 1':
        [24,1,9,29,92,79,4,90,42,35],
        'Production Day 2':
        [2,43,17,5,31,89,44,49,34,84]
    })
df = pd.DataFrame(data)
print(df)
# Without FactoryZone, it works:
df = df.drop(['Factory Zone'], axis=1)
image = df.plot(kind="bar")

And the data looks like this:

  Unnamed: 0 FactoryZone       Factory Name  Production Day 1  Production Day 2
0           1    AMERICAS  Chocolate Factory                24                43
1           2    AMERICAS     Crayon Factory                 1                17
2           3        EMEA           Pet Shop                 9                 5
3           4        EMEA     Bonbon Factory                29                31
4           5        APAC           Lil Grey                92                89
5           6    AMERICAS         Jobs Ur Us                79                44
6           7        APAC          Toys R Us                 4                49
7           8        EMEA          Carrefour                90                34
8           9    AMERICAS       Gibberish US                42                84
9          10        APAC          Food Inc.                35                62

Answer 1

You can create this plot by first creating a MultiIndex for your hierarchical dataset where level 0 is the Factory Zone and level 1 is the Factory Name :

import numpy as np                 # v 1.19.2
import pandas as pd                # v 1.1.3
import matplotlib.pyplot as plt    # v 3.3.2

df = pd.DataFrame(
    {'Factory Zone': ['AMERICAS', 'AMERICAS', 'AMERICAS', 'AMERICAS', 'APAC',
                      'APAC', 'APAC', 'EMEA', 'EMEA', 'EMEA'],
     'Factory Name': ['Chocolate Factory', 'Crayon Factory', 'Jobs Ur Us',
                      'Gibberish US', 'Lil Grey', 'Toys R Us', 'Food Inc.',
                      'Pet Shop', 'Bonbon Factory','Carrefour'],
     'Production Day 1': [24,1,9,29,92,79,4,90,42,35],
     'Production Day 2': [2,43,17,5,31,89,44,49,34,84]
    })

df.set_index(['Factory Zone', 'Factory Name'], inplace=True)

df

#                                   Production Day 1  Production Day 2
#  Factory Zone       Factory Name      
#      AMERICAS  Chocolate Factory                24                 2
#                   Crayon Factory                 1                43
#                       Jobs Ur Us                 9                17
#                     Gibberish US                29                 5
#          APAC           Lil Grey                92                31
#                        Toys R Us                79                89
#                        Food Inc.                 4                44
#         EMEA            Pet Shop                90                49
#                   Bonbon Factory                42                34
#                        Carrefour                35                84

Like Quang Hoang has proposed, you can create a subplot for each zone and stick them together. The width of each subplot must be corrected according to the number of factories by using the width_ratios argument in the gridspec_kw dictionary so that all the columns have the same width. Then there are limitless formatting choices to make.

In the following example, I choose to show separation lines only between zones by using the minor tick marks for this purpose. Also, because the figure width is limited here to 10 inches only, I rewrite the longer labels on two lines.

# Create figure with a subplot for each factory zone with a relative width
# proportionate to the number of factories
zones = df.index.levels[0]
nplots = zones.size
plots_width_ratios = [df.xs(zone).index.size for zone in zones]
fig, axes = plt.subplots(nrows=1, ncols=nplots, sharey=True, figsize=(10, 4),
                         gridspec_kw = dict(width_ratios=plots_width_ratios, wspace=0))

# Loop through array of axes to create grouped bar chart for each factory zone
alpha = 0.3 # used for grid lines, bottom spine and separation lines between zones
for zone, ax in zip(zones, axes):
    # Create bar chart with grid lines and no spines except bottom one
    df.xs(zone).plot.bar(ax=ax, legend=None, zorder=2)
    ax.grid(axis='y', zorder=1, color='black', alpha=alpha)
    for spine in ['top', 'left', 'right']:
        ax.spines[spine].set_visible(False)
    ax.spines['bottom'].set_alpha(alpha)
    
    # Set and place x labels for factory zones
    ax.set_xlabel(zone)
    ax.xaxis.set_label_coords(x=0.5, y=-0.2)
    
    # Format major tick labels for factory names: note that because this figure is
    # only about 10 inches wide, I choose to rewrite the long names on two lines.
    ticklabels = [name.replace(' ', '\n') if len(name) > 10 else name
                  for name in df.xs(zone).index]
    ax.set_xticklabels(ticklabels, rotation=0, ha='center')
    ax.tick_params(axis='both', length=0, pad=7)
    
    # Set and format minor tick marks for separation lines between zones: note
    # that except for the first subplot, only the right tick mark is drawn to avoid
    # duplicate overlapping lines so that when an alpha different from 1 is chosen
    # (like in this example) all the lines look the same
    if ax.is_first_col():
        ax.set_xticks([*ax.get_xlim()], minor=True)
    else:
        ax.set_xticks([ax.get_xlim()[1]], minor=True)
    ax.tick_params(which='minor', length=55, width=0.8, color=[0, 0, 0, alpha])

# Add legend using the labels and handles from the last subplot
fig.legend(*ax.get_legend_handles_labels(), frameon=False, loc=(0.08, 0.77))

fig.suptitle('Production Quantity by Zone and Factory on both days', y=1.02, size=14);

---

References: the answer by Quang Hoang, this answer by gyx-hh

Answer 2

An idea that gives a close plot is to plot to each Factory Zone in a subplot that are place next to each other:

# setting up the subplots
fig, axes = plt.subplots(1, len(df['Factory Zone'].unique()), 
                         figsize=(12,4),
                         sharex=True, sharey=True, 
                         gridspec_kw={'wspace':0},
                         subplot_kw={'frameon':False})

# use groupby to loop through the `Factory Zone`
for (k,d), ax in zip(df.groupby('Factory Zone'), axes):

    # plot the data into subplot
    d.plot.bar(x='Factory Name', ax=ax)
    
    # set label to the `Factory Zone`
    ax.set_xlabel(k)
    
    # remove the extra legend in each subplot
    legend = ax.legend()
    handlers = ax.get_legend_handles_labels()
    ax.legend().remove()
    ax.grid(True, axis='y')

# reinstall the last legend
ax.legend(*handlers)

Output:

Answer 3

The solution offered by Patrick FitzGerald has a single line that was deprecated in Matplotlib 3.4 and will be removed in 2 minor releases. (I'd put this as a comment rather than an answer, but I don't have enough reputation yet!)

Change:

if ax.is_first_col():

to

if ax.get_subplotspec().is_first_col():