The Visual Display of Quantitative Information
For the last couple of years I have been in search of theories in Data Visualization. Educate myself on the fundamentals. My search has taken me to many books and blogs. But none as remarkable as Edward Tufte book seminal work on the subject. This is a short refresher of the core concepts. Even as I write for myself, it may be of some use to a passing busy programmer.
- Graphical Excellence: that which gives a viewer maximum ideas in shortest time with least ink in the smallest space
- Graphical excellence is nearly always multivariate. Charts depicting behavior of two variables with respect to each other are always more insightful than simple time-series or progression graphs
- ‘Graphical Integrity’ reigns supreme. Beware of distortions. Thre representation of numbers as physically measured on the surface of the graphic itself should be directly proportional to the numerical quantities represented (as an aside, this book might be a good read on distortions!)
- The number of information carrying (variable) dimensions depicted should not exceed the number of dimensions in the data. Beware of area charts depicting single variable variations
- Maximize the data-ink ratio, within reason. Erase non-data-ink, within reason. Revise. Rethink.
- Moire vibrations in statistical charts are chartjunk. Gridlines, often, are chartjunk. 3D, often, is chartjunk. More than 3 colors are, often, chartjunk. Piecharts are always chartjunk. Easy graphing software is leading to more chartjunk and more amazing chartjunk
- Awesome examples of clarity by revision - redesigning boxplots, barcharts and my personal favorite - the super intuitive dot-dash plot combining marginal distribution with a bivariate distribution!
- Use coordinates and axes with thought - maximize data-ink
- Organize and order the flow of graphical information presented to the eye - charts should intelligently use what are known facts on cognitive abilities of human brain
- Balance and optimize data-density = (number of data entries)/(area of the graphic). Try to maximize it. Else shrink the graphic
- Pay attention to line weights
- Curious case of the Golden Rectangle
- And, finally, it was John Tukey who once said - there is no data that can be displayed in a pie chart, that cannot be displayed BETTER in some other type of chart.
comments powered by Disqus