Predictive Analytics in Healthcare: It's Not Happening
Predictive Analytics Key Component of Customer Experience Management
Predictive Analytics in Higher Education
Trends in Computational Healthcare Research and Digital Therapeutics
Increasing Your Bottom Line with Predictive Analytics
Rich Jolly, VP Analytics & Strategy, Avamere Family of Companies
Data Lake: Building a Bridge between Technology and Business
Shaun Rankin, Ex SVP Data Management, Citizens Bank
Leveraging Predictive Analytics to Power Customer Data &...
Dean Abbott, Co-founder & Chief Data Scientist, SmarterHQ
Thank you for Subscribing to CIO Applications Weekly Brief
Lies, Damn Lies, and Data Visualization
By Douglas Duncan, CIO, Columbia Insurance Group
Your truth, my truth, the objective truth….philosophers, politicians, and lawyers could talk all day about what it means, but the end result is that it, like beauty, is in the eye of the beholder. This is not a satisfying conclusion. One could wish human nature were different. Solipsists may have it right in that we can only know what we experience.
However, my personal experience tells me that I need to make a living, and that you do too. If your living is in providing information to others in a meaningful and consumable manner, then you need to think hard about how you present that information and what version of the truth you are trying to convey.
All data presented beyond a listing of the raw numbers comes in the form of visualization. Whether you present a structured table, a simple graph, a complex 3-D model or an intricate infographic, you are helping the consumer visualize the data. Every choice you make, purposefully or by accident, is important for how the data are depicted: font, color, weight, choice of graphical elements, composition, scaling, legend or accompanying text. What you do not show graphically is as critical as what you do show. Even the story behind the data is vital – do you buy into the data premise and origin, or do you dismiss it as hokum and highly suspicious?
The advantages of providing visualizations to help understand data are undisputed.
Ask yourself four key questions when you begin to design a data visualization to make your point or share your information:
1. Do the data themselves present a clear picture or is a visualization needed? If you are trying to show that the cost of IT services goes up as the number of users increases, you may not need a fancy visualization. Avoid gilding the lily and stick to listing the few relevant facts if that is the point you are trying to make. Not everything needs to be in a chart to be meaningful.
While a picture is worth a thousand words, a chart is worth even more in data points
2. Does the visualization give the essence of the data? Nobody wants to reverse-engineer a graphic in order to understand the underlying point. If your goal is to highlight your aging infrastructure, only bring in other elements, such as location or hardware brand, when it is a meaningful attribute of the case for change.
3. Does this depiction of the facts leave anything open to interpretation? When your goal is to share information and not make a specific case, then let your data visualization speak what it will. In cases where you are trying to make a point and support it with data, be sure your visualization makes it obvious. Avoid falling in love with your own work. Double and triple check that the visualization illustrating the data has a clear message and that the truth it conveys is not just your pet view. Do not expect the consumer of the data to be able to make nuanced evaluation of the picture you have painted. Paint your picture decisively and with style!
4. Could you create a visualization with the same data that supports a very different viewpoint? The link between the underlying data and the visualization should be clear and it should be one-way. Do not try to convince with a pretty picture alone… the data themselves should be very compelling; the visualization only makes the point more clear.
IT remains an art. No organization has the same setup as any other. Technology practitioners must be both artists and scientists. For data visualization, this is doubly needed. Do not assume that the graph auto-generated by your tool actually sends the message you want to convey. Think carefully about the need for the visualization, the essence of the data supporting it, how it could be interpreted and whether there are alternative conclusions that the data supports. You have license to be creative, which is necessary for you to successfully convey your truth through data visualization.