Here at Duke, I do data visualization. That covers everything from helping people clean up their data to teaching workshops on software and design techniques, and even to advertising innovative local visualization work. I see a lot of unrealized opportunity in the intersection between research computing and data visualization, and I’m looking forward to sharing my ideas on this blog periodically.
Most of the work I do and see with visualizations tends to focus on the ability of visualizations to summarize research findings and communicate them to an audience. And don’t get me wrong; communication is a wonderful use of data visualization. What better way to give people a quick overview of a trend you’ve uncovered than by a well-structured bar chart or scatter plot?
But many people have never considered using visualization to help themselves uncover that trend. Visualization for exploration and analysis can be a very powerful complement to other research methods. You can use a series of scatter plots to discover which variables have the strongest relationships, or make a quick map to verify that your geocoding results are accurate. Being able to access individual data points without aggregation or summarization can improve our understanding of distributions, relationships, clusters, outliers, variance, etc.
I encourage you to consider incorporating exploratory data visualization into your larger research computing toolkit. Iterating between standard data processing or analysis techniques and visualizations can help verify your results or even suggest new areas and methods of analysis. Even the process of thinking of how to visualize something may inspire a new algorithm or data structure.
Of course, just make sure you give me a buzz when you have questions or interesting results!
Angela Zoss, firstname.lastname@example.org
Data Visualization Coordinator, Duke University Libraries/OIT