There was a discussion in the comments to the Graph Makeover Contest concerning the use of slope plots and parallel coordinate plots when we are not plotting against time. I wrote to Alfred Inselberg, author of the book Parallel Coordinates: Visual Multidimensional Geometry and Its Applications, asking if he wanted to join the conversation. Alfred responded by plotting the expenditures data in the second part of the contest with parallel coordinates plots. Figure 1 shows his figure, which adds another visualization to the eight different ways to plot the table of consumer expenditures that are described in Visualizing the Table of the Graph Makeover Contest.
His message said, “Due to the big difference in scale between values and their percentage change (e.g., 56050, -41) it is better to display the data in TWO plots.”
Figure 2 shows the plot of the percent changes.
- Figure 2. Parallel coordinates plot of percent change of expenditures. Source: Alfred Inselberg Click to enlarge
Alfred’s message continued to say:
For the actual values twelve variables parallel axes appear: Annual-Expenditure, Food, Food-(at)-home, Food-away-from-home, Housing, Apparel, Transportation, Gas, Healthcare, Health-Insurance, Entertainment, Insurance-pensions, and two categorical axes: Year, Owners-(value)1_Renters-(value)0 . Note that the first axis has a larger maximum (scale) than the remainder in order to accommodate the difference in values. Some sample values are shown. The Owners data is in GREEN while the Renters data is in BLUE. Note that there were greater changes for the renters in some variables on the left but this reversed for variables shown on the right. For the most part trends/proportions were the same for renters and owners for both years. Note that owners and renters spent more for Housing in 2010, for Food-away and Apparel than in 1986.
For the percentage changes the same twelve variables axes are shown as in previous plot but only one categorical axis on the right with Owners – 1 and Renters 0 is needed. Again some sample values are shown with GREEN for owners’ data and BLUE for the renters.
Please comment on the article here: Effective Graphs