It’s been a while since our last dispatch. At Plotly, we’ve been heads down working on features for both Dash open-source and Dash Enterprise. In case you don’t subscribe to Plotly’s Twitter or Medium, I thought I’d round up some of my favorite headlines.
(If you want to automatically receive Dash Club newsletters, sign up!)
Version Check
Dash OSS is on v1.16.3 and Dash Enterprise is on v4.0.1. Get upgrading, y’all!
Dash for Julia
With collaborators at Julia Computing, we’ve officially launched v1.0 of the Dash Julia library, adding the project to our list of Long Term Support codebases. We hope that Dash for Julia will be what Shiny has been for R, Application Designer has been for MATLAB, and Dash has been for Python. Sign up for the free Dash for Julia intro Webinar tomorrow with @alexcjohnson!
Dash for Autonomous Vehicles
With collaborators at Lyft and NVIDIA, we’ve open-sourced 3 Python Dash apps for AV research and data. While mainly prototypes, these apps illustrate the rapid development approach of Dash versus a full-stack team approach. And they’re super fun to play with! Get a free guided tour of these Dash AV apps on Tuesday, October 27, by Plotly’s lead ML engineer @xhlu.
Tips for high-performance choropleth maps in Dash
In recent community news, I love this hot tip from community powerhouse @Emil: Use Shapely and/or geobuff to compress geojson polygons to load massive choropleth maps quickly in Dash. Wish I had thought of this!
Data Science Workspaces
At Plotly, we’ve switched from our desktop code editors to developing Dash apps and ML models inside Dash Enterprise’s Data Science Workspaces. It’s a faster way for us to onboard new engineers developing Dash apps and Jupyter notebooks and crucially, dev = prod. Like all Dash Enterprise features, Workspaces are open-core, based on the excellent Theia project from the Eclipse Foundation . If you’d like to take Workspaces for a spin or see a demo, please get in touch.
Jobs at Plotly!
Plotly hasn’t let the pandemic slow down our mission to open-source the communications layer of scientific computing. We continue to hire in 2020. There are currently 3 positions open in Canada, including one for full-time open-source development (JavaScript/Typescript).
GitHub Sponsors
Related to Jobs at Plotly, we’ve enrolled Plotly’s GitHub organization in GitHub Sponsors. If you use and enjoy Plotly’s open-source libraries, but are not a Dash Enterprise customer, you now have a way to show your appreciation and augment our work. Contributions from GitHub Sponsors will go directly to expanding Plotly’s staff of developers In Canada who are dedicated full-time to open-source development and issue solving.
Dash Open Source Releases
We’re currently on Dash v1.16.3. My favorite features in the last 4 months: Revamped callback graph with detailed performance metrics, Dash Deck (WebGL & 3D Maps), performance improvements, JupyterDash to write Dash apps in Jupyter Notebooks, and Gantt charts with px.timeline(). We release a new Dash OSS release every month, so see the announcements for more.
From the Archives
UI Revision to maintain the user’s graph selections & zoom level in between callback updates, sunburst charts for multi-dimensional & hierarchical data, Datashader plots for visualizing >1M data points, pattern-matching callbacks that update an arbitrary number of dynamic components, DataTable conditional formatting, and Plotly Express.
Dash for OpenAI’s GPT-3 API
In case you missed them, these open-source GPT-3 Dash apps are insane. Ask for a chart that you want in a sentence, and GPT-3 will generate it for you with unsettling accuracy. This is some Westworld shizzah. Special thanks to OpenAI’s CTO for getting us early access to the GPT-3 API.
Dash integration with scikit-image
With collaborators at Scikit-image and the Chan Zuckerberg Institute like @Emmanuelle, we’ve been working hard to augment interactive image analysis and computer vision in Python. Scikit-image docs now include interactive Plotly graphs for next-level interactivity with image features. Also, see the annotating image traces with shapes examples.
Show & Tell
Thanks to everyone who has been sharing their apps & components in the forum’s show & tell! So much good stuff in here. Some of my recent favorites:
Components:
- Dash Leaflet by @Emil
- Download component by @Emil
- Lottie animations by @Emil
- Transcript (Transpiling Python to JavaScript) by @Emil
- Server-side caching by @Emil
- Large File Uploader by @fohrloop
- Video engine by Skyratlnd
Apps:
- mRNA explorer by Atharva Arya for Vaccine Degradation Prediction and more COVID apps
- Financial analysis toolbox by @NAVA
- Nuclear fusion reactor simulations by @plasmapotential
- XGBoost & SVM Telco Customer Churn Predictor by @cepel
- gnucash Ledger Explorer by @saufrec
- Encarta-Inspired World Atlas by @dan_baker
Coming Soon…
We’ve got some awesome work coming down the pipe before 2020 mercifully ends:
- Our Chief Scientist, Jon Mease (@jmmease) , has been hard at work integrating Dash & Holoviews for single-line dashboard crossfilter in Python
- In collaboration with Microsoft Azure and bioinformatics researchers at the University of Kaiserslautern, we’re laying the foundations for “Dash.NET” - an F# implementation of Dash. An alpha version of the library is already installable from nuget and
Dash.NETapps can be deployed to Heroku or Dash Enterprise. Documentation will be available in early 2021. - Azure & AWS installation screencasts for Dash Enterprise are in the works. By popular demand, we aim to streamline the installation of Dash Enterprise on these clouds through screencasts, improved public documentation, and ever-deepening Kubernetes support.
- Dash Enterprise 4.1 will be out later this Fall. Stay tuned! If you haven’t upgraded to Dash Enterprise 4.0 yet, get in touch with your customer support team.
Stay safe & healthy out there —
Chris