Why we have no easy way to visualize data in Augmented Reality? — Now we do!

Data Visualization in Augmented Reality (or AR) made easy via iPhone and Python is now possible with the iOS App Augmented Data.

Photo by 丁亦然 on Unsplash

With the advance of the new iOS camera and technologies like LiDAR focused specifically in improving AR experiences, how can the data science community take advantage of these innovations in a quick and easy way? The main barrier for the exploration of AR in data science seems to be the technical costs of plotting a single data point in a mobile application. Mobile development paradigm, a new programming language and slow iteration between plots are some of the reasons that kept a data scientist away from AR technologies.

Thinking to solve this technology gap I developed an iOS app that reads inputs generated by python code and plots its data in a tridimensional space embedded in AR. This app is called Augmented Data and is available for free on iOS.

Using the app, anyone can see the plots available in its workspaces, and via a standard iPython notebook, a data scientist can change the plots in the workspaces to quickly iterate over ideas and insights. See some of the plots below:

A circumference with metallic dots on the left and a Multidimensional Gaussian distribution on the right. Screenshots taken from Augmented Data App.

Finally, bellow you can see me updating the AR data in realtime with a few lines of Python code:

A gif of me updating AR data in realtime using a Jupyter notebook (right) and the phone screen (left).

For the machine learning aficionados, the app also counts with hand-segmentation so you can ‘touch’ the data. I trained my own MobileNetV2 model and deployed in the App so you can play around and interact with the data.

Conclusion

I believe that one of the reasons that AR hasn’t been used much for data visualization is due to its high initial toll of having to learn mobile development along side another language (like Swift). However, the Augmented Data can shorten this gap serving as an interface between languages like Python or R and the mobile environment, in this way the data scientist can focus in analysing and iterating in its plots efficiently.

The Augmented Data App solves three problems:

1. No need to learn another language (like Swift) just to plot your data in Augmented Reality.
2. No need to compile and deploy code every time you want to make a change in an AR plot.
3. App is integrated with Python, allowing data scientist to focus on insights not in technical challenges.

I will continue exploring with AR and the new App to create visualizations in this new design space.

What do you think of AR for data science?

More?

If you want more dope data visualisation checkout these articles

• An introduction about Bayesian Rule — Go here
• Do Warm or Cold Filters in Your Pictures Drive More Clicks? — Go here

Follow me on Twitter: https://twitter.com/solvingthehuman