Sylvain Corlay

---

---

Jupyter widgets are powerful tools for building user interfaces with graphical controls such as sliders and text boxes inside a Jupyter notebook. Interactive widgets can also be rendered in Sphinx documentation, nbviewer, and static web pages. Jupyter widgets are more than a collection of controls, they also are a framework that makes it easy to build custom GUI controls. Examples of custom widget packages include libraries for interactive 2-D charting (bqplot), 3-D graphics (pythreejs, ipyvolume), mapping (ipyleaflet), and more.

1. This tutorial begins with an overview of the functionalities provided by the base package:

• the collection of interactions available in ipywidgets
• the styling and layout abilities
• the embedding of widgets in contexts other than the notebook
• the use of the @interact automatic GUI generation tool.

We walk the audience through the main functionalities of the core package showing how it can improve their workflow for both the exploratory phase and the presentation of their work.

2. More than a collection of simple controls, Jupyter interactive widgets have become a rich ecosystem of visualization libraries from 2-D and 3-D plotting to mapping, CAD and molecular visualization. We then make a brief presentation of some of the popular widget libraries built upon Jupyter widgets:

• 2-D charting (bqplot)
• 3-D visualization (pythreejs, ipyvolume)
• mapping (ipyleaflet).

We guide the audience in the creation of a dashboard assembled from simple controls and visualizations in the Jupyter notebook using both core and third-party widgets.

3. In the last part of the tutorial, we show how the Jupyter interactive widgets system can be extended with new controls.

We start with a brief example of developing a custom widget including both Python and JavaScript, detailing some aspects of the front-end infrastructure. We explain how the new widget can be packaged into an installable python library using the widget-cookiecutter template project.

Hands-on session talk: 100 minutes long, code heavy, practical session