Uber’s amazing geospatial data viz tool joins the open source community


Imagine a geospatial data visualization tool that doesn’t require coding, works in the browser, doesn’t need to be installed, can render millions of GPS points in the blink of an eye, and is super-easy to use. That’s Uber’s kepler.gl geospatial tool for you in a nutshell.

Uber has always recognized location data as one of its biggest assets. It is what has governed the decision-making process across the board at the ride-hailing startup. Uber has used analytics of location big data to understand how many trips start at a particular location, determine the most convenient pickup points for its riders, find out the fastest routes, and even see how many drivers make interstate trips.

Naturally, Uber’s need for fast exploration of geospatial big data led to the birth of kepler.gl. Shan He, the architect behind kepler.gl, describes the platform as, “a data-agnostic, high-performance web-based application for visual exploration of large-scale geolocation data sets. Built on top of deck.gl, kepler.gl can render millions of points representing thousands of trips and perform spatial aggregations on the fly.”

Last week, Uber made the kepler.gl toolbox open source. It comes preloaded with Mapbox and all you’ve got to do is drop in a CSV or a JSON file to have the tool up and running in seconds. It’s been reported to be even easier than Tableau by early users, and you can see some of the beautiful visualization types it would allow you to explore below:

kepler.gl’s point, arc, and heatmap layers (top) and grid, hexbin, and polygon layers (below) provide rich geospatial data analysis

Uber explains that all layer geometry calculations are GPU-accelerated, which has enabled a smooth rendering of millions of points and made kepler.gl a much more powerful web tool than traditional cartography software. There are plenty of filter options for you to play around with, including adding time playbacks to visualize spatiotemporal data, excluding outliers using histograms, and refining data to a smaller set for comparison.

Below, you can see a sample origin-destination map created by the team at Uber using kepler.gl. Leveraging the commute data of residents in England and Wales, this map uses bi-color arcs to connect the residences (in yellow) and workplaces (in magenta).

We are hoping to see some really cool maps come out of this platform. You can get started on your first one here.