Ctrl+K
This website is for version 5. You can find the documentation for version 4 here.

Vega-Altair

Roadmap#

This page contains the current roadmap for the Vega-Altair project. The roadmap is informed by the needs of the Vega-Altair community, and the priorities of the active project contributors. It’s designed to communicate the direction of the project, but it’s not a commitment that these items will be completed in a particular timeframe. If you would like to help contribute to any of these areas, or suggest new ones, please start a discussion.

Vega-Altair is deeply integrated with other components in the Vega ecosystem, and as such many items on the roadmap will require work in other projects. Abbreviations for these projects are included at the end of project descriptions where relevant:

API Ergonomics#

The primary job of the Vega-Altair library is to provide an ergonomic Python API for generating Vega-Lite JSON specifications, and there are several improvements here that we would like to investigate.

Areas of focus:

  • Improve the syntax for creating condition expressions with two or more predicates.

  • Explore the possibility of a new operator, *, for modular compositing of sub components. See also Deneb.jl by @brucala.

  • Standardize the API of methods that convert charts into other formats (alt.Chart().to_<format>).

  • Released in 5.2: Add type hints to the public API and most of the internals so that users can type check their Altair code with a static type checker such as mypy. This will also make it easier for other packages to integrate with Altair.

Documentation#

We want to continue to improve Vega-Altair’s official documentation to be more useful for both beginning and experienced users.

Areas of focus:

  • Incorporate a conceptual guide that includes best practices of effective data communication.

  • Update and extend the tutorial section and replace outdated materials.

  • Add usage guide with best practices for publishing Vega-Altair charts in various contexts including websites, research papers, embedded dashboards, and interactive platforms.

  • Add guide on building domain specific visualization packages on top of Vega-Altair. For example, Vega-Altair for soccer analytics.

  • Add documentation for the expression language that is available throughout the API.

Ecosystem Integration#

We want Vega-Altair to be well integrated with the PyData ecosystem. It should work well with popular libraries and ecosystem standards.

Areas of focus:

  • Provide integration with the broader Python DataFrame ecosystem (beyond pandas). Ensure that all of Vega-Altair’s features are available to any DataFrame library that implements the DataFrame Interchange Protocol.

  • Work with dashboard toolkit maintainers to ensure that Vega-Altair is well supported. Write documentation for best practices for making Vega-Altair’s interactive features available in Python.

Gridded Data Support#

Vega-Altair currently requires tidy tabular data as input, so it is not currently a natural choice for working with gridded data. We would like to extend the project to include native support for gridded datasets.

Areas of focus:

  • Add support for Python array/tensor interchange protocol (through the __dlpack__ interface)

  • Add support for creating charts from Xarray DataArray objects (rendering large arrays may require the performance work described elsewhere).

Increased Coverage of Statistical Charts#

While Vega-Altair includes support for many types of statistical visualizations, there are a few important types that are not yet possible.

Areas of focus:

Map Tile Support#

We want Vega-Altair to provide first-class support for displaying map tiles from xyz tile providers like OpenStreetMap. We’ve released a first version of altair_tiles to accomplish this. Feedback is very welcome!

Scale/Performance Improvements#

In the traditional Vega-Altair architecture, a chart’s entire input dataset is serialized to JSON and transferred to the browser for data transformation and rendering. Rendering itself is then performed by The Vega JavaScript library using the Canvas API (which is not GPU accelerated). This architecture has many advantages (e.g. chart specifications are fully self-contained and portable to Python-free rendering environments), but it is not well suited for creating charts of large datasets.

Through a variety of enhancements, our goal is to allow all Vega-Altair charts to scale comfortably to over 1 million rows.

Areas of focus:

  • Provide optional integration with VegaFusion to automatically move data transformation steps from the browser to efficient multi-threaded implementations in the Python kernel.

  • Utilize binary serialization of datasets in Apache Arrow IPC format between the Python kernel and the browser. This will significant reduce serialization time for large unaggregated visualizations such as scatter plots.

  • Support creating Vega-Altair charts that reference tables in external database systems, and convert data transformation steps to SQL that can be evaluated by the database before the results are transferred to the Python kernel (VF).

  • Add support for GPU accelerated rendering. This will enable rendering of large unnagregated visualizations at interactive speeds. For example, pan and zoom interactions on a large scatter plot (VG).

Static Image Export#

Even though Vega-Altair charts are rendered by the Vega JavaScript library, it’s important to provide reliable (and easy to install) support for exporting charts to static images. Image export was dramatically simplified in Vega-Altair 5.0 with the adoption of VlConvert, which has no external dependencies on a system web-browser or Node.js installation. Now that image export is easy to install and easy to use, we want to improve support for publication workflows.

Areas of focus:

  • Released in 5.1: Support configurable pixel density in PNG image export (VC).

  • Released in 5.2: Support exporting charts to vector PDF files with embedded text (VC).

On this page