Data Analysis in Virtual Reality

Carl Anderson
3 min readOct 21, 2016

--

I’ve been doing data analysis for quite a few years but this week I had an entirely novel experience: data analysis in virtual reality.

I was attending a conference put on by Looker, a BI tool provider. During one of their internal hackdays, one of their engineers had hacked together “LookVR,” a combination of an HTC VIVE headset, some visualization libraries, and the Looker API as a data source.

Donning the headset, you enter a basic 3D world surrounded by oceans and mountains but below you was a wooden stage. Using a pair of hand controllers, you were able to use a menu to select a number of standard data visualizations, such as a 3D scatter chart, bar chart, and a word cloud.

Making a selection, the visualization rose up from the ground like a Phoenix and appeared in front of you. You could then use the controllers to manipulate the visualization. For instance, with a 3D scatterplot you could expand the size so that you were within the cloud or points and then you could rotate the visualization around you. Look up, and you see the points above you. Look down, and you see the points below. You were in the point cloud, exploring the relationships among the variables.

Inside the interactive word cloud

Similarly with a word cloud. As a small speck in the data, you could turn yourself or you could turn the cloud around you. Look up and see a word of interest, you could use the controller to click on a word and window popped up with metadata. The controllers were surprisingly easy to use. The fellow conference attendees and I were able to pick up how they worked in about 10 seconds.

About to dive into the skyscrapers

Interestingly, in a 3D barchart, you could scale up the chart so that the bars become skyscrapers soaring above. You could stroll through the chart like it was New York, which was appropriate as that is exactly where we were.

Scale, rotation, and metadata were not the only interactivity. You could interact with the axes of these visualizations and slice into the data, reducing the dimensions. Now you had a one dimensional barchart that you could scale, translate, and rotate.

This was an incredible immersive experience that is hard to describe. Yes, you can create a 3D scatterplot in Excel or R, and make it rotate, but this was an entirely different experience.

It was built mostly for fun, a bit of a gimmick, but ultimately a chance to show off the possibilities of their API. I’m not claiming that VR data analysis will become mainstream — although Microsoft and Facebook are making big bets on the general technology — but interacting with the 3D scatterplot was a pretty compelling use case. I could imagine interacting with a 3D Gephi social network plot or an Ayasdi topological data analysis visualization. It might have a place within very specific, likely more scientific than business, applications.

Thank you Looker and especially Wil Gieseler. You provided us with a fascinating, fun, thought-provoking, and trippy and experience.

--

--

Carl Anderson

Director, Data Science, Indigo Ag. Author of "Creating a Data-Driven Organization" (2015, O'Reilly). Web: carlanderson.ai