Over 100 people packed into IBM’s conference room this morning for the Big Data Seminar: What Does All This Data Mean?

Kicked off with an overview by moderator Richard Dale, where he defined big data, identified where data comes from, reviewed ways in which to cull meaning from it. With the question: ‘Is Visualization Art or Science?’, panel members, Irene Greif of IBM, Martin Leach of The Broad Institute, and Andrew Pandre of Sears Holdings Corporation, dove right in.

View the presentation here. 

The panelists agreed that visualization is more than just providing the customer or end-user with a chart or graph explaining the data. It provides a tool to get people to act on the data, making decisions and implementing processes. It allows them to have a sense of ownership of what otherwise is just a lot of information that would have been useless.

Andrew provocatively stated that 80% of visualization is the data, 10% is the story behind the data, and only 10% is actually visualization. But it is that last 10% that provides the user with meaning. Furthering that sentiment, was Martin who said that visualization is a data reduction tool.

Irene described visualization as a way to get people drawn in to a conversation, allowing them to get involved and collaborate. She continued to say visualization should be used as a tool to create hypotheses and decide whether further analyses should be done, but not necessarily to make a final decision.

Siting sparklines as the simplest form of visualization, overall panelists agreed that the simpler the visualization can be made, the better. Because the human eye can only take in so much information, good visualization should only show small amounts of information, but be dynamic, allowing the opportunity to drill down further. Other attributes should be ease of understanding, flexibility, and utilizing colors and shapes. And in the end the most important part of visualization is understanding what you are hoping to accomplish at the onset.

Finally, Richard asked the panelists to name their favorite visualization tools, all agreed the answer is dependent upon what they are trying to show, but the list included: ClickView, Many Eyes, Spotfire, Tableau.