Sergey Nivens - Fotolia
Any time a representative of car sharing service Uber Technology Inc. shows up at an analytics conference, his or her session is always packed.
People crowd into the room for two reasons. First, Uber does a lot of interesting things with advanced analytics, and getting a peak under the hood at how it all works can inspire new projects at other enterprises.
But the second reason people like Uber presentations is that they are always visually appealing. The company has invested heavily in developing an open source data visualization tool, and this makes for great production value.
It's not just about pretty pictures though. Uber uses its software to help users develop new insights from data, and the results can be impressive.
In November 2016, the company made its data visualization software, called deck.gl, open source. This month, it introduced an updated version that the company thinks will help users get more out of their data visualizations.
"I'm a very big advocate for open source, so I think that open source can help the business in many other ways," said Nicolas Garcia Belmonte, head of data visualization at Uber.
He said the decision to make deck.gl open source is a two-way street. He hopes that visualization developers outside of Uber will find interesting ways to use the open source data visualization tool and derive value from it. This innovation will ideally feed back into how Uber uses deck.gl, eventually leading to new ways the company can visualize its own data. In this way, the company hopes to gain by helping others.
This software is particularly strong at visualizing geospatial data. Uber has used it to visualize rider pickups to learn things about the rider experience, such as where riders and drivers frequently miss each other.
But it's not all about geospatial data. Data science teams at Uber are also using the software to visualize machine learning models to get a more intuitive sense of how they are working. In this way, the tool is being used to optimize algorithms used in the UberEATS food delivery service to improve delivery time estimates.
Deck.gl is tuned to handle large data volumes, which makes sense given the huge data sets Uber developed it to visualize. It works by applying the concept of layers to geospatial data. Users can start with a basic map with some data points, and then layer on visualizations that depict changes in the data over time or other aspects of the data.
This month's update to the open source data visualization tool included several new layers. One set of updates expanded the number of ways geospatial data could be visualized, and another set of layers made it possible to visualize non-geospatial data.
Garcia Belmonte said he hopes more businesses use the software to develop interesting ways of looking at data.
"I'm eager to see how developers use deck.gl, which can inspire us."
Expanded feature set makes data visualization software more complicated
Focus on business value is key to implementing data visualization tools
Keep it simple to get the most out of data visualization