How to Display Pandas DataFrame As a Heatmap   

1. Overview

In this tutorial, we'll learn how to display Pandas DataFrame as a heatmap.

So we might start with: what is a heatmap in Data Science? According to wikipedia:

A heat map (or heatmap) is a data visualization technique that shows the magnitude of a phenomenon as color in two dimensions.

2. Setup

We are going to create test DataFrame following two articles:

• How to Create a Pandas DataFrame of Random Integers
• How to Easily Create Dummy DataFrame with Test Data?

We are using np.random.randn(10, 3) to create DataFrame with 10 rows and 3 columns- with random values:

import numpy as np
import pandas as pd
from pandas.util.testing import makePeriodSeries

s = makeTimeSeries(10)

cols = ['col_1', 'col_2']
df = pd.DataFrame(abs(np.random.randn(10, 2)), columns=cols)
df['item'] = 'item: ' + df.index.astype(str)
df['date'] = s.index

3. Pandas: Display DataFrame as heatmap with style.background_gradient

Pandas offer method style.background_gradient() which helps us very easily to create beautiful colored heatmap:

df.style.background_gradient(cmap='Greens')

The background gradient it will applied only for the numeric columns:

The method background_gradient() take as argument cmap which can have different values like:

• Blues
• Greens

To learn more about Pandas colors and palettes please visit:

• How to Get a List of N Different Colors and Names in Python/Pandas
• Choosing Colormaps in Matplotlib

4. Seaborn: Display DataFrame as heatmap with sns.heatmap

There is a library for data visualization called Seaborn: statistical data visualization.

This library offers method called: seaborn.heatmap()

The method works only on numerical values. So we can use it as follow:

import seaborn as sns

sns.heatmap(df[['col_1', 'col_2']])

the DataFrame will looks like:

If you try to call the method: sns.heatmap() on the whole DataFrame you will get an error:

ValueError: could not convert string to float: 'item: 0'

Another way to solve the error is by pivoting data on some columns. For example we can pivot on columns:

• "date"
• "item"

and get as values column - "col_1":

sns.heatmap(df.pivot("date", "item",values='col_1'))

This will convert the DataFrame into beautiful heatmap:

Again we can provide parameter cmap which can take similar values as the background_gradient().

5. Interactive heatmap with Plotly

If you like to make your DataFrame as aa interactive heatmap then you can use library called:

• Plotly: The front end for ML and data science models

Again as Seaborn we need to use only numeric values:

import plotly.express as px
fig = px.imshow(df[['col_1', 'col_2']])
fig.show()

Otherwise errors will be raised. The resulted heatmap will looks like:

For categorical data we can use pivot() or similar operation in order to make it good for plotting as a heatmap.

The error is a bit different:

TypeError: Object of type Period is not JSON serializable

6. Conclusion

We covered the most popular ways to convert DataFrame to:

• heatmap for numeric and non numeric data
• heatmap with seaborn
• data transformation for categorical data with pivot
• interactive heatmap

This article will help you to select the best way to present your numeric data.