Adding percentage of count to a stacked bar chart in plotly

Question

Given the following chart created in plotly.

I want to add the percentage values of each count for M and F categories inside each block.

The code used to generate this plot.

arr = np.array([
        ['Dog', 'M'], ['Dog', 'M'], ['Dog', 'F'], ['Dog', 'F'],
        ['Cat', 'F'], ['Cat', 'F'], ['Cat', 'F'], ['Cat', 'M'],
        ['Fox', 'M'], ['Fox', 'M'], ['Fox', 'M'], ['Fox', 'F'],
        ['Dog', 'F'], ['Dog', 'F'], ['Cat', 'F'], ['Dog', 'M']
    ])

df = pd.DataFrame(arr, columns=['A', 'G'])

fig = px.histogram(df, x="A", color='G', barmode="stack")
fig.update_layout(height=400, width=800)

fig.show()

Census

Answer 1

As far as I know histograms in Plotly don't have a text attribute. But you could generate the bar chart yourself and then add the percentage via the text attribute.

import numpy as np
import pandas as pd
import plotly.express as px

arr = np.array([
        ['Dog', 'M'], ['Dog', 'M'], ['Dog', 'F'], ['Dog', 'F'],
        ['Cat', 'F'], ['Cat', 'F'], ['Cat', 'F'], ['Cat', 'M'],
        ['Fox', 'M'], ['Fox', 'M'], ['Fox', 'M'], ['Fox', 'F'],
        ['Dog', 'F'], ['Dog', 'F'], ['Cat', 'F'], ['Dog', 'M']
    ])

df = pd.DataFrame(arr, columns=['A', 'G'])

df_g = df.groupby(['A', 'G']).size().reset_index()
df_g['percentage'] = df.groupby(['A', 'G']).size().groupby(level=0).apply(lambda x: 100 * x / float(x.sum())).values
df_g.columns = ['A', 'G', 'Counts', 'Percentage']

px.bar(df_g, x='A', y=['Counts'], color='G', text=df_g['Percentage'].apply(lambda x: '{0:1.2f}%'.format(x)))

Pantin

The code looks good, just that for me the last line had to be changed to: px.bar(df_g, x='A', y='Counts', color='G', text=df_g['Percentage'].apply(lambda x: '{0:1.2f}%'.format(x))) – Truelove

Answer 2

Note that you can now specify plotly barnorm and text_auto arguments to achieve this. Have a look at your example:

# Libraries
import numpy as np
import pandas as pd
import plotly.express as px

# Data
arr = np.array([
    ['Dog', 'M'], ['Dog', 'M'], ['Dog', 'F'], ['Dog', 'F'],
    ['Cat', 'F'], ['Cat', 'F'], ['Cat', 'F'], ['Cat', 'M'],
    ['Fox', 'M'], ['Fox', 'M'], ['Fox', 'M'], ['Fox', 'F'],
    ['Dog', 'F'], ['Dog', 'F'], ['Cat', 'F'], ['Dog', 'M']
])

df = pd.DataFrame(arr, columns=['A', 'G'])

#Plotly Code
fig = go.Figure()

fig = px.histogram (  df,
                      x="A",
                      color="G",
                      barnorm = "percent",
                      text_auto= True,
                      color_discrete_sequence=["mediumvioletred", "seagreen"],
                ) \
        .update_layout (

                    title={
                            "text": "Percent :A - G",
                            "x": 0.5
                          },

                    yaxis_title="Percent"
                ) \
        .update_xaxes(categoryorder='total descending')

fig.show()

In general it should be the prefered solution over calculating the percentages yourself. Here the output:

Calculating

TypeError: histogram() got an unexpected keyword argument 'text_auto' – Tarsia

Make sure you have most recent update for your plotly installation. – Calculating

Upgrading from plotly 5.4 to 5.6 allowed me to use 'text_auto' – Carlina

That should be the accepted answer as in March 2022. – Pantin

Answer 3

You can also use the text argument in px.bar() to specify custom text labels for each bar. If you want to display the values as percentages, you can use the str.format() method to format the text labels.

topCom = df["Make"].value_counts().sort_values(ascending=False).head(10)
px.bar(x= topCom.index ,y= topCom/742*100,
text=[f"{value:.2f}%" for value in topCom/742*100],
color_discrete_sequence=['#4C4C6D'])

Gorrono