Using pandas crosstab to create a bar plot

Question

I am trying to create a stacked barplot in seaborn with my dataframe.

I have first generated a crosstab table in pandas like so:

pd.crosstab(df['Period'], df['Mark'])

which returns:

  Mark            False  True  
Period BASELINE    583    132
       WEEK 12     721      0 
       WEEK 24     589    132 
       WEEK 4      721      0

I would like to use seaborn to create a stacked barplot for congruence, ans this is what I have used for the rest of my graphs. I have struggled to do this however as I am unable to index the crosstab.

I have been able to make the plot I want in pandas using .plot.barh(stacked=True) but no luck with seaborn. Any ideas how i can do this?

Hexapod

As an FYI, stacked bars are not the best option because they can make it difficult to compare bar values and can easily be misinterpreted. The purpose of a visualization is to present data in an easily understood format; make sure the message is clear. Side-by-side bars are often a better option. Stacked bars may be appropriate for comparing total amount across groups, or for comparing relative differences between quantities in each group. Stacked Bar Graph. – Cacomistle

Answer 1

• As you said you can use pandas to create the stacked bar plot. The argument that you want to have a "seaborn plot" is irrelevant, since every seaborn plot and every pandas plot are in the end simply matplotlib objects, as the plotting tools of both libraries are merely matplotlib wrappers.
• Here's a complete solution (using the data creation from @andrew_reece's answer).
• Tested in python 3.8.11, pandas 1.3.2, matplotlib 3.4.3, seaborn 0.11.2

import numpy as np 
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

n = 500
np.random.seed(365)
mark = np.random.choice([True, False], n)
periods = np.random.choice(['BASELINE', 'WEEK 12', 'WEEK 24', 'WEEK 4'], n)

df = pd.DataFrame({'mark': mark, 'period': periods})
ct = pd.crosstab(df.period, df.mark)
    
ax = ct.plot(kind='bar', stacked=True, rot=0)
ax.legend(title='mark', bbox_to_anchor=(1, 1.02), loc='upper left')

# add annotations if desired
for c in ax.containers:
    
    # set the bar label
    ax.bar_label(c, label_type='center')

Oedipus

Answer 2

• The guy who created Seaborn doesn't like stacked bar charts (but that link has a hack which uses Seaborn + Matplotlib to make them anyway).
• If you're willing to accept a grouped bar chart instead of a stacked one, following are two approaches
• Tested in python 3.8.11, pandas 1.3.2, matplotlib 3.4.3, seaborn 0.11.2

# first some sample data
import numpy as np 
import pandas as pd
import seaborn as sns

N = 1000
np.random.seed(365)
mark = np.random.choice([True, False], N)
periods = np.random.choice(['BASELINE', 'WEEK 12', 'WEEK 24', 'WEEK 4'], N)

df = pd.DataFrame({'mark':mark,'period':periods})
ct = pd.crosstab(df.period, df.mark)

mark      False  True
period               
BASELINE    124   126
WEEK 12     102   118
WEEK 24     118   133
WEEK 4      140   139

# now stack and reset
stacked = ct.stack().reset_index().rename(columns={0:'value'})

# plot grouped bar chart
p = sns.barplot(x=stacked.period, y=stacked.value, hue=stacked.mark, order=['BASELINE', 'WEEK 4', 'WEEK 12', 'WEEK 24'])
sns.move_legend(p, bbox_to_anchor=(1, 1.02), loc='upper left')

• The point of using pandas.crosstab is to get the counts per group, however this can be bypassed by passing the original dataframe, df, to seaborn.countplot

ax = sns.countplot(data=df, x='period', hue='mark', order=['BASELINE', 'WEEK 4', 'WEEK 12', 'WEEK 24'])
sns.move_legend(ax, bbox_to_anchor=(1, 1.02), loc='upper left')

for c in ax.containers:
    
    # set the bar label
    ax.bar_label(c, label_type='center')

Bebe

Answer 3

I recommend using seaborn histplot, especially if you have more than 2 values for the categorical data.

import numpy as np 
import pandas as pd
import seaborn as sns

N = 1000
np.random.seed(365)
mark = np.random.choice([True, False], N)
periods = np.random.choice(['BASELINE', 'WEEK 12', 'WEEK 24', 'WEEK 4'], N)


df = pd.DataFrame({'mark':mark,'period':periods})
group_df = df.groupby(['period', 'mark']).size().reset_index(name='count')

plt.figure(figsize=(10,6))
ax = sns.histplot(group_df, y='period', hue='mark', weights='count',
             multiple='stack', palette='colorblind')

Gisellegish