The Data Visualisation Catalogue Blog

VR Data Visualization: Useful For Multidimensional Data?

Continuing on at looking into the claims being made about virtual reality (VR) data visualization, we will now look at the second most commonly made claim:

VR data visualization is great for multidimensional data analysis; VR allows you to view the data in ‘multiple dimensions’

From taking a guess, we can assume that this relates to VR’s ability to display data in 3D. So the additional dimension here would be depth (z-axis). But because we don’t perceive depth well, due the distortions and other issues that can occur, would having an extra dimension be of much benefit?

Also, 2D charts can already visualise multidimensional and multivariate data with Bubble Charts, Chernoff Faces, Parallel Coordinates Plots, Radar Charts. So what would VR do better?

If VR is capable of displaying more than one extra dimension, how would this be beneficial? In Stephen Few’s critical post on VR data visualization, he makes the point that:

[…]And let us not forget that we can only hold from three to four chunks of visual information in working memory at once, so even if VR could add many more dimensions of data in some way, it would be of no use to our limited brains if we weren’t able to process all of those dimensions simultaneously.

~ Can VR Enhance Data Visualization?, Perceptual Edge

However, from my research online, this is what I’ve heard being said about the VR’s ability to visualize multidimensional data:

Okay, let us talk about living in the real world again. There are five senses that determine our existence in the real world, but in a typical scenario, we could only use our sight to analyse and interpret data. But what if we could use other senses too? This is exactly what Virtual Reality offers in data visualization. For example, take the case of hearing. With data-audio relationships, we could easily determine the location, subject, and significance of a specific data through its direction, loudness, and type! With haptic feedback gloves gaining popularity, we are not far away from a period where we could actually feel the data.

~ 5 Amazing Advantages of Virtual Reality in Data Visualization, SRUSHTIVIZ

By using multiple senses, we can enhance our ability to process data with more dimensions. While it may be a bit radical to talk about taste and smell in a data visualisation context, it is not outside the bounds of possibility to ‘feel’ data. This is technically achievable right now with haptic feedback gloves.

~ 5 Reasons to Use Virtual Reality for Data Visualisation, Towards Data Science

Before VR made its entry, users had only their eyes to analyze the data, but post VR, multi-dimensional data analysis is possible. And that means, not just the use of hands, but hearing as well. This enables them to understand the subject, location and significance of a particular data source.

Maybe it is a little out of bounds to talk about feel and taste in data visualization, but perhaps, such a day is not too far behind, right?

~ 6 Ways Virtual Reality will Transform Big Data Visualizations, Cabot Technology Solution

So from all this we can see that the claim of VR allowing you to view the data in ‘multiple dimensions’ doesn’t just relate to depth or another form of visual communication. VR also allows you to process data in the form of sound, touch, smell and soon in the near future, even taste.

Maybe we’ll be able to tell if the data smells bad?

Still, I don’t see how this would aid in data analysis. But in terms of communicating the data and a narrative, I could see this being really novel and engaging. Not everyone learns well visually and are instead better at absorbing information through different means, for example, by auditory or kinesthetic learning.

To get some industry insight into the topic, I spoke with Suzanne Borders from BadVR. Here’s what she had to say:

Visualizing univariate data in 2D is pretty straightforward but issues begin arising when the number of dimensions in the data increases. These issues occur because most data visualizations are bound by the 2D visual displays used to present them.

Additionally and unfortunately, as modern datasets trend towards ever greater scale and complexity, it becomes more and more difficult over time to visualize multivariate datasets in a way the accurately communicates all the existing depth and complexity.

Human analysis on 2D screens require multiple different summary charts, graphs, and reports which must be manually cobbled together by humans to inform decision-making. Obviously, this method has it’s downsides. Complex interrelationships are impossible to find, and due to sheer scale, a large amount of the dataset remains unexamined from certain angles.

If you remove the compromises required by 2D screens, a whole slew of new ways to visualize data become possible. Users can view one big picture of their entire dataset and literally ‘step inside their data’ to see and analyze in real-time. VR also offers the ability to massively increase the scale of data presented to the user, allowing users to see everything all at once, removing the need for summary reports or charts.

Abstracting large datasets down isn’t necessary when you have depth, position in space, position to other data points, texture, animation, and many more ways to communicate multiple data attributes, and a limitless environment in which to do so.

I found this article from Evan Warfel which attempts to explain how VR is useful for high-dimensional analysis: Everything Wrong with Traditional Data Visualization and How VR is Poised to Fix It.

From there I discovered a nice example which demonstrates where VR can be useful in multidimensional data visualization:

There’s also really cool video from Project NEO by Francois Bertrand, which demonstrates VR’s usefulness in viewing multidimensional data to detect patterns. There’s even a feature to view the data in a 6-dimensional hypercube:

In an article titled New insights into the suitability of the third dimension for visualizing multivariate/multidimensional data: A study based on loss of quality quantification, there is some scientific evidence to suggest the superiority of using 3D visualization over 2D when it comes to visualizing multidimensional and multivariate data.

Although this study doesn’t include VR technology, it does show how that even using 3D visualization on a screen yields better results on particular visual tasks. For example, when it came to distance perception, the error values produced in the 3D version are much lower than those in 2D (p. 13 – 14). Overall, it was concluded that there was a significant loss in quality when switching from a 3D to 2D visualization.

. . .

In summary, it’s pretty obvious that VR allows you to visualise multidimensional data, not only by using depth and animation, but also in non-visual forms of communication such as sound, touch and smell. While all of these features are not unique to VR, I would also argue that the combination of all the other features into an immersive experience is the unique trait of VR. While there seems to be some evidence to suggest that VR is useful in the display and analysis of multidimensional data, more research needs to be done.

I’d also like to give a special thanks to Andrea Bravo for the academic suggestions and insight.

Related Reading

New insights into the suitability of the third dimension for visualizing multivariate/multidimensional data: A study based on loss of quality quantification, A. Gracia, S. González, V. Robles, E. Menasalvas and T. von Landesberger

DataViz: visualization of high-dimensional data in virtual reality, E. Feng and X. Ge

Virtual Reality-based Human-Data Interaction, E. A. Widjojo, W. Chinthammit and U. Engelke

Other VR posts

Data Visualization Technology Virtual Reality

Next post:

Further Exploration #13: 3D Treemap Spheres and Cylinders

Blog Home

Previous post:

Further Exploration #12: 3D Treemap Cubes