Let’s talk statistics...
Did you know that Twitter records 6,000 “tweets” every second? That equates to 350,000 tweets sent per minute, a whopping 500 million tweets per day and around 200 billion tweets per year. Did you know that Google instigates 40,000 search queries every second on average, leading to 1.2 trillion searches per year worldwide? Or that Visible Equity performs analysis on over 21.1 million current, unique loans, which equates to an overall balance of about $340B and 20+ million borrowers. It is no understatement that we truly live in the age of big data.
You’ve heard all about the hype of “big data”, but what does it mean for you? Hal Varian, Google’s chief economist said, “The ability to take data – to be able to understand it, to process it, to extract value from it, to visualize it, to communicate it – that’s going to be a hugely important skill in the next decades.”
Data visualization is not new. In fact, the need to communicate and present data has been around for hundreds if not thousands of years. What is new, however, is the contemporary appetite for and interest in the intersection of art and science; that there is a difference between looking at data, and truly seeing it. The greatest value of a picture is when it forces us to notice what we never expected to see. Through data visualization, we are seeking to portray data in a new light, one that allows a process of discovery and ultimately expands stories hidden behind the data’s raw state.
As the information age accelerates forward, our roles are changing. More and more of us are becoming responsible for the analysis, presentation and interpretation of data. What once was a specialist’s job now seeps into every aspect of the digital world. Thus we might ask ourselves, “Why do I design visualizations in the way I do?” Maybe you fall into the following categories:
- You have a certain design style based on personal taste
- You just play around until something emerges that you instinctively like the look of
- You trust software defaults and don’t go beyond that in terms of modifying the design
- You have limited software capabilities, so you don’t know how to modify a design
- You just do as the boss tells you – “can you make me some fancy charts?”
It’s very possible that you have never even thought about your visualization design approach until now. Don’t worry; you’re not alone! Feeling like a fish out of water in the world of ‘big data’ is as common as the phrase itself is overused. This article will equip you with the tools necessary to communicate what your data are saying. Because technology and preferred avenues differ greatly, we will focus primarily on the methodology behind data visualization – placing emphasis on the concepting, reasoning and decision-making involved.
Keys to Visualize your Data
1. Seek to develop a visual sixth sense — To successfully interpret visualizations, and use them to explore and present complex information, we first need to train our eyes and brains to unweave graphics with little or no conscious effort. Reading visualizations is just like anything else; The more you practice, the better you’ll be. Seek out good graphics and study them. Even the most creative people are borrowers from their predecessors.
2. Strive for form and function —This is the classic debate of style versus substance. Although that balance may be difficult to achieve, our aim should be to hit that sweet spot where something is aesthetically inviting and functionally effective. The following image is taken from an animated wind map. It is a beautiful piece of work, exceptionally well designed and executed, but it also serves its purpose as a way of informing users about the wind patterns, strength, and directions occurring across the United States.
For many beginners, our general advice is to initially focus on the functional aspects of your visualization. Ensure that the foundation – the function – supports the necessary data before enhancing and embellishing. Over time you will be much more confident and capable of uniting the demands in unison.
3. Justify the selection of everything you do - Ensure that your visualization is a deliberate design.
The inclusion, exclusion, and execution of every single mark, characteristic, and design feature must be done for a reason. You should be prepared to challenge everything; the use of a shape, the selection of a color pallet, the position of a label, or the use of an interaction. It is also important to remember that any visual aspect not directly related to data representation should only be included to aid the appearance, not hinder it.
4. Focus on Accessibility – In the words of Edward Tufte, Professor of Statistics at Yale University, “overload, clutter, and confusion are not attributes of information, they are failures of design.” Ultimately readers want your design to do what it should do — inform. That is never to say you must simplify, dilute or reduce the essence of the subject matter. Rather, the complexity of the visualization should be in direct proportion to the complexity of the data. This might mean something is not immediately easy to interpret, and that’s okay. Difficult ideas will be hard to digest at first. Effective approaches to visualizing data ensure that the efforts engaged in learning how to read the visual eventual leads the user to understand the visual.
For example, we visualize Instagram photos taken during Hurricane Sandy in Brooklyn, NY and detect behavioral patterns during emergency times as they are reflected in users’ photos. Below is a radial plot visualization of 23,581 photos taken during 24 hours in Brooklyn area during hurricane Sandy. The photos are organized by time (angle) and darkness of the photo itself (distance from the center of the graph). Note the line marking a change in the number of photos and their brightness, corresponding to the moment of the power outage in the area. This sudden and dramatic visual change reflects well the intensity of the human experience during the event.
Radial plot visualization of 24 hours from the Brooklyn area during hurricane Sandy
5. Never deceive the receiver. There are such things as visualization ethics that relate to the potential for deception, whether intentionally or otherwise, from an ineffective and inappropriate representation of data. One particular example is the following graphic, which seems to show the opposite of what’s really going on:
At first glance, it looks like gun deaths are on the decline in Florida. But a closer look shows that the y-axis is upside-down, with zero at the top and the maximum value at the bottom. As gun deaths increase, the line slopes downward, violating a well-established convention that y-values increase as we move up the page. Obeying visualization ethics should always be an objective for any project. Be sure that your designs are clear, concise and convey a sense of respect for your readers.
Another example of data deception often comes in the familiar form of the pie chart. These ghastly graphics are very difficult for the human brain to properly interpret. Although they can be good for comparisons of two to three different data points with very different amounts of information, more often than not the pie chart is non-informative, or even misleading. For example, suppose that the following pie charts show a distribution of votes for five political candidates at three points in time, A, B, and C.
If you have a hard time picking up on a pattern, you’re not alone. Your brain just won’t easily detect it. However, viewing the exact same data in bar format, the signal is much clearer.
See? Much clearer. We can see obvious trends at each time interval, as well really understand the size of one group to relative to the others.
Let’s take a look at one more example. Below we see a breakdown of the European Parliament Party portrayed in two different styles of pie charts.
At first glance, it would appear that of the angled pie chart shows the EPP and S&D are roughly the same size. Upon further scrutiny of the standard, 2-dimensional chart, however, we see this is anything but the truth (not to mention the 3-D version prevents us from effectively comparing any of the Parliament parties).
The bottom line is pie charts often fail to portray what they are designed for – making information visually informative. In most cases, it is probably a better idea to use a bar chart in lieu of a pie chart.
Putting it all Together
Popular methods of data visualization that still dominate boardrooms across the financial world include the line and bar charts. Did you know these methods originate from the eighteenth century? Let’s look at a simple example. We have extracted data from a ‘dummy’ financial institution in Visible Equity’s very own backyard – Salt Lake City. Below we have a basic bar chart comparing the geographic distribution of the institution’s new and used auto portfolios throughout the state.
Although the data are crisp and informative, the chart seems to fall on a flat note. Putting in practice the tools now in hand, we can enhance our newfound visual sixth sense and focus on the form, function, accessibility and aesthetics of what our data are really trying to portray.
The following example shows statewide geographic distribution:
We now focus in on our primary target lending area, Salt Lake City.
Having looked at various examples of quality data visualizations, the hope is that you are truly beginning to see these stories come to life. Learning to create effective, simple, and beautiful designs can benefit anyone, from financial analysts to CEO’s. In this millennial age of “big data” being able to visually communicate the story of your data is not a handy skill, it is a necessity.
After all, if you are going to say it, say it well.