Around 40,000 years ago, the first man-made two-dimensional symbols appeared on cave walls. These are among the earliest known visual expressions of human ideas. Today, we continue the tradition of visual communication through an incredible range of media - paints on a canvas, film, pottery, computer pixels, photography, and countless others.
With a recent rise in availability of massive datasets, so too has risen an acute public interest in understanding what stories these large datasets are telling, and in visualizing these insights in useful and compelling ways. The data visualization industry is booming.
At Correlate.io, we love diving into datasets, understanding the stories they tell, and finding practical, yet beautiful and engaging ways to share those stories through our data visualization platform. Transforming raw datasets into visual narrative helps us understand relationships, correlations, and stories in our data in a way that we may not immediately be able to see in looking at raw datasets.
The data visualization process is a curious beast that begins with seemingly incomprehensible volumes of information, moves through an intensive data cleaning, classification, statistical and mathematical modeling, analysis, design process, and ends with a visualization. The entire process of building a visualization is a complex interdisciplinary melange of domains. One day, we took a step back and watched as our team of diverse professionals worked on developing a visualization. We were amazed at just how many different perspectives were necessary to create the dynamic end goal.
In hopes of satisfying our attempts to pinpoint what exactly data visualization is, we put our staff to the test. We asked them what they thought - is data visualization an art? Or is it a science?
What the IB5k team says...
“It’s definitely a combination of both. Visualizations are often the result of a sequence of analytics operations applied over a data set. The goal of the visualization is to communicate the pattern that those operations found within the collection of data. For a visualization to be effective, it has to be engaging. It has to interact with human beings so that it can communicate in a compelling and clear way the patterns that are found, and that part is a lot of art. There is some science about the psychology of visualization that informs how that’s done, but it’s still a lot of art.”
- Ezra Lee (Data Scientist, IB5k and Correlate)
“I think the ability to create meaning out of data is very much an art. It takes creativity and ingenuity to come up with these graphs and be able to apply meaning and infer real-world things from simple data. I feel like data is your palette, and the visualization is the painting. Sure you can throw a bunch of data into a plot and get something out of it, but to create something meaningful that other people can understand and appreciate and admire takes a real artist.”
- Jackson Blankstein (Developer, IB5k and Correlate)
“I think it’s a science. It doesn’t matter how pretty it looks, it doesn’t matter how clear the fonts are or the differentiation that you want to show in the visualization. If the data set and the analysis aren’t done right and correct and rigorous, it’s just a picture. I don’t think it’s art, it’s science.”
- Jung Lee (Product Specialist, IB5k and Correlate)
“Data visualization is both an art and a science. Data science requires modeling, testing, and mathematical validation. However, how one chooses to display this information requires finesse and attention to both aesthetic and technical detail in order to correctly, beautifully and simply visualize the data. Data can be hugely misrepresented due to a lack of understanding or an over-simplified aesthetic treatment. Data visualization requires both the understanding of the raw data and the ability to use graphical elements wisely and accurately to depict actionable insights from that raw data.”
One of my favorite quotes of all time is, "Lies, damned lies, and statistics," by Mark Twain. I think it's important that we are aware that statistical analysis is a field subject to our predilections; i.e. data can be manipulated to allow its author to tease out desired results. Thus, we must be vigilant in analyzing our data and we must be deft authors of design, using our capacity as artists and scientists to help unearth the underlying, objective truths in our data sets.
- Laura Seach (Project Manager, IB5k)
“I think it’s an art. When data scientists either collect or receive data, they’re responsible for telling a story through a rigorous process of data cleaning, statistical analysis, modeling, and so much more. And what is art but storytelling? Like choreography is a way of telling a story through movement, like a painting can tell a story through strokes on a canvas - same with music. It’s all about storytelling. From that standpoint, it’s an art.”
- Angela Schöpke (Project Manager, IB5k and Correlate)
“I think that effective data visualization is both art and science. By that I mean the following: Clearly it is a science, there is hard data involved and there are scientific ways to analyze it, structure it and interpret that data. One must abide by certain rules during the stages of collecting and aggregating data as well as applying rigorous mathematical tools to look for patterns that are scientifically meaningful.
That said, there is an artistic aspect not just to the presentation of the data (dashed or dotted lines, color choices, patterns and shapes and so on) which must clearly communicate the information, conclusions and patterns, but also in the initial stages when one is confronted with a massive dataset and one must often find a place to begin. Knowing where to begin is itself part art, having a feel for the real-world situation which generated the data and having an intuition for how things might be related in order to even begin the search and analysis.”
- Justin Wyss-Gallifent (Developer, IB5k and Correlate)
What do you think? Is data visualization an art or a science? Send us a note with your thoughts!