Share Your Jupyter Notebook Charts on the Web
PixieDust now lets you turn charts into standalone web pages
In my last blog post, I talked about effortless operationalization of data science with the new PixieGateway server — part of PixieDust — that lets you deploy Jupyter Notebook PixieApps as standalone web apps.
Starting with PixieDust 1.1.1, the PixieGateway server now supports sharing of charts created with the display() function, effectively turning them into standalone web pages. This new feature is extremely attractive because there is absolutely no coding necessary.
Simply use the interactive PixieDust display() function from within a Jupyter Notebook to build your chart, click on the Share button, add an optional description and voila, you get a URL that you can distribute to your team. They can then see the chart without needing to run a Jupyter Notebook. All they need is a web browser.
Note: PixieGateway is still in alpha. Much like the ability to publish PixieApps as interactive web apps (release 1.1), the ability to publish charts as web pages (release 1.1.1) also uses the PixieGateway server. Thus, this feature is still in alpha and therefore not ready to be used in production.
Chart sharing in action
Using display() to create a chart, you’ll now see a Share button next to “Options”:
The Share dialog will ask for the following information:
- PixieGateway server: If you don’t have one, see this blog post for instructions on how to set one up.
- Description: Optional free-form text explaining what the chart is about. It will be added to the standalone webpage.
Click Share, and if all goes well, you should get a URL for your chart webpage that is ready to share:
You can copy the URL and/or click on the link to go directly to the page:
Improving this feature
We have multiple ideas on how to improve this feature:
- Security: Password-protect your charts.
- Embed: Provide embed code so you can include the chart in your own websites (similar to YouTube).
- Auto-refresh: Schedule automatic refresh of charts, with options like “Refresh every x hour” or “Every day at 10 p.m.”
This last idea would be a useful feature if the data source used to build the chart were to change over time (e.g., sales data). In this case, the chart would become live without the need for you to re-run the notebook. Even better, you could add a live chart to your website if you were to combine it with the embed feature proposed above.
As always, we need your feedback and ideas to make PixieGateway server better. If you like this feature and, like me, think that it will help unlock the value of data science beyond Jupyter Notebooks, then join the conversation on GitHub by creating an issue. Pull requests are most welcome!
