Noah Iliinsky, from IBM's Center for Advanced Visualization, discusses the power of visualizations and the reasons why they're a necessity to fully understand what's 'inside the data'. In this post, Noah will dive into these reasons, including why spreadsheets and written text are simply not the tell-all that visualizations can be.
You've got Big Data; that's a good start. However, unless you've got a good way to extract knowledge from it, you've got Big Nothing.
So what's the right path?
Well, when it comes to analysis and presentation of data (of any size), few approaches are as valuable and powerful as visualization. Here are some reasons why.
Visualization is high-bandwidth.
The largest data channel going into your brain is your optic nerve. One study conducted by the University of Pennsylvania estimated the bandwidth of the optic nerve at around 8 mbit/second, almost as fast as first-generation ethernet or wifi. That’s a huge amount of data. A well designed visualization can leverage this bandwidth to communicate a lot of knowledge very quickly.
Visualization is persuasive.
According to a joint study from Dartmouth College and the University of Exeter, there’s some evidence that visualizations are more effective at changing people’s minds than words alone. Our brains are built for visual perception and our visual system is extremely sophisticated at pattern recognition, edge-finding, shape recognition, and related tasks at both large and small scales. We also parse and understand visual input very quickly. And we’re much better at quickly extracting meaning from visualizations than from digits or tables.
Visualization exposes patterns.
There is value in patterns. Almost everything we consider “interesting” about data is either a pattern, or a pattern violation. When it comes to data, patterns are visible as trends, gaps, and outliers. Once we understand a pattern or trend, we understand (or “see”) what’s going on with a system. Given that understanding, any change in the trend or any violation of that pattern, is interesting. Visualization is a fantastic way to see, understand, and share these interesting characteristics of data.
Visualization is excellent for overviews.
As your data becomes larger, spreadsheets will become overwhelming, but a well-designed visualization can represent what’s going on in the big picture (if not all of the details at once). A fantastic example of this is below, from the Gapminder project.
(Note: The visualization is free material from Gapminder.)
This visualization shows life expectancy (y-axis), income (x-axis), region (color), and population (size), for over 200 countries. It also highlights change over time for two countries. This would be an overwhelming amount of data for a spreadsheet, but is made accessible by this visualization.
Visualizations can be responsive.
Visualizations can be designed to reveal specific detail that you care about, and not distract you with unrelated information. And visualizations can be used to extract levels of hierarchy or aggregation, revealing more detail as you zoom in.
Visualizations aren't always the right choice.
Of course, there are times when text and numbers are superior to visualizations. When you want to ensure a specific message is communicated, you'd write that explanation out in text. Another example of when visualizations aren't appropriate would be if you needed relatively high precision in your values, showing the values with digits instead. Or, if you need to communicate a huge amount of raw data, a spreadsheet would be more accessible than thousands of graphs.
But, if you want to find or communicate trends or outliers, show an overview, allow interactive exploration, and be persuasive, visualizations are hard to beat.
Visualizations & IBM's RAVE engine.
To harness the power of visualization in a standard and accessible way, IBM has built the RAVE engine. RAVE is a display technology capable of rendering visualizations from the simple to the complex. It's being built into IBM's analytics tools, including the Cognos family. It is also the underlying visualization technology used by all of Watson's visual analytics tools. There is a selection of free visualizations and RAVE-enabled analytics tools downloadable from the AnalyticsZone.
Visualizations are one of the most effective ways to extract knowledge from data, and powerful tools make creating the right visualizations possible. Check out all of our visualization-related resources, including white papers, tools, and examples all on the AnalyticsZone Marketplace and within AnalyticsZone's Visualization community.