UC Visualization Day

typescript

Earlier this year I attended Visualization Day at the University of Cincinnati. Speakers this day focused their talks on how to make graphs, charts and other visual representations of data easier to read, interpret and act upon. The keynote speaker was Stephen Few, author of Information Dashboard Design, widely considered the bible of good dashboard and visualization design techniques. Stephen identified 7 types of quantitative relationships with recommendations for each:
• Time series (values over time) – use line charts, easy to visualize changes over time
• Ranking – rank highest to lowest or vice versa, easier to compare values across items
• Part to whole – don’t use pie charts, people can’t visualize area differences easily
• Deviation – variances shown as bars or lines above or below a reference line
• Distribution – visually see distribution of values across a histogram (i.e. bell curve)
• Correlation – relationship between 2 values, use scatter plots with trend lines
• Geospatial – use when location of values is important
Stephen also noted that it is important to identify the basis for comparison in any charts you create. As visualization pioneer Edward Tufte once stated, you need to ask yourself “compared to what” when you are comparing values in charts.

The next speaker was Jeff Shaffer who is the VP of IT and Analytics at Unifund. Jeff echoed many of Stephen’s comments about why pie charts are bad. Additionally, Jeff listed the 4 C’s of visualizations:
• Clear – easily seen
• Clean – thorough, complete
• Concise – brief but clear, simple
• Captivating – draws you in
Jeff also suggested use of several websites to help with visualization design. They are:
• stagewww.vischeck.org – checks views for color blind readability
• stagewww.juiceanalytics.com/
• vis.stanford.edu/
• stagewww.edwardtufte.com

The last speaker of the day was Andy Walter who is the VP of Business Intelligence at P&G. While much of Andy’s talk focused on changes within P&G that improved visualization techniques and usage, Andy noted several good points for creating successful integration of visualization techniques into an organization. They identified several key skills for analysts creating visualizations within P&G:
• Analytics background – math, engineering, etc…
• Domain knowledge of business area – key to understanding information
• Effective communication skills – can verbalize what is needed
The key findings resulting from their transformation at P&G were “present the data even if you can’t drill down right away” and “don’t wait until the data is perfect before presenting it to the business”.  Very good goals to live by in the data world…