DS23: Intro to Data Viz 101 and a makeover!

by Joselito Bondoc

At the end of the first day of Data Viz 101 with Coach Andy, he gave us a task to complete this week's #MakeoverMonday.

Figure a. Original viz by @owenlhjphillips

What I like

  • Clear and clean design and layout
  • Good title and subtitle to set context
  • arrows and annotations help audience read the chart
  • gridlines faded to the background
  • great colour use

What could be improved

  • could remove row gridlines
  • foul type labels can be placed somewhere less intrusive
  • adding more jitter to the dot plots would allow easier selection of referees

Overall I already quite like this chart, there's not a whole lot I would change. And my Makeover is just me making those 3 little changes and making a slightly similar version of this already great visualisation.

Figure b. My own makeover.

My changes
During Data Viz 101, Andy taught us how to approach an analysis of a new data set. he told us to ask simple questions like:

  • When
  • What
  • Where
  • Who
  • How
  • Why

I started with asking: how did tendencies of calling fouls change over time?; what fouls were called more often? and how do referees compare to each other? In the end, I wanted to focus on the latter because I wanted to stay true to the original visualisation (and I wanted to try making a parallel coordinates chart).

As we can see in Figure b., the parallel coordinates chart serves a very similar purpose to the dot plots in the original viz.y-axis is a min-max normalised value for each foul type, where a higher value indicates a higher average number of fouls called per game. This was done because each foul type had a different range of numbers and would have made the chart quite cluttered. There is a drop-down menu in the top-right to highlight a specific referee throughout the viz.

At the bottom, we can see a dot plot with a jitter to allow for more spacing between each dot vertically, which makes it easier to see the distribution compared to a dot plot with no jitter or a parallel coordinate chart. Moreover, the x-axis here is not normalised and shows the average number of fouls called per game. A reference line indicating the season average is also included as a point of comparison. The labels for the foul types also now have their own space in both charts.

Overall, I enjoyed making a parallel coordinate chart and I think it works quite well, especially if you normalise the values and highlight specific lines (whilst pushing the others in the background).

Thumbnail photo by Andre Tan.