Boxplot

A boxplot summarizes the distribution of a numeric variable for one or several groups. It allows to quickly get the median, quartiles and outliers but also hides the dataset individual data points. In python, boxplots are most of time done thanks to the boxplot function of the Seaborn library. Boxplots are also described in the online course.

⏱ Quick start

Seaborn is definitely the best library to quickly build a boxplot. It offers a dedicated boxplot() function that roughly works as follows:🔥

Basic boxplot with Python and Seaborn from various data input formats.

# library & dataset
import seaborn as sns
df = sns.load_dataset('iris')

sns.boxplot( x=df["species"], y=df["sepal_length"] )

⚠️ Mind the boxplot

A boxplot is an awesome way to summarize the distribution of a variable. However it hides the real distribution and the sample size. Check the 3 charts below that are based on the exact same dataset.

To read more about this, visit data-to-viz.com that has a dedicated article.

Basic boxplot. You can quickly read the median, quartiles and outliers of each group.

If you add individual points with jitter, a bimodal distribution appears for group B

If you have a very large dataset, the violin plot is a better alternative than jittering

Boxplots with `Seaborn`

Seaborn is a python library allowing to make better charts easily. The boxplot function should get you started in minutes. The examples below aim at showcasing the various possibilities this function offers.

Let's start basic. The most simple boxplot, based on 3 differents input formats

Everything you need concerning color customization on your boxplot: transparency, palette in use, manual control..

Learn how to highlight a specific group in the dataset to make your point more obvious

If you have group and subgroups, you can build a grouped boxplot

Control the order of groups in the boxplot. It makes the chart more insightful

To avoid hiding information, you can add individual data points with jitter

Since individual data points are hidden, it is a good practice to show the sample size under each box

Customization: border width

Customization: use notch

Customization: box width

If you have both groups and subgroups, you'll be interested in a grouped violin plot

🔎 `violin` function parameters→ see full doc

widthbwcolor

string→ color under the curve

From the web

The web is full of astonishing charts made by awesome bloggers, (often using R). The Python graph gallery tries to display (or translate from R) some of the best creations and explain how their source code works. If you want to display your work here, please drop me a word or even better, submit a Pull Request!

A combination of a violin plot and a boxplot. Allows the comparison of several groups with statistical test results on top.

Contact

👋 This document is a work by Yan Holtz. Any feedback is highly encouraged. You can fill an issue on Github, drop me a message onTwitter, or send an email pasting yan.holtz.data with gmail.com.

GITHUB TWITTER