« Sore-thumb graphics | Main | Spinning multi-color 2 »

Comments

a.c.

I don't understand what's going on in the junkart version. There doesn't seem to be near enough grey in each bar.

a.c.

Oh, I guess I see it now. I guess that's proportion of each cell's contribution to the total emissions, not to the column (gas)'s emissions.

Harlan

a.c.: yeah, but I think a better way to do it would be to make the width of each column proportional to the column's percentage of the whole. So the CO2 bars would be 8 times wider than the NO2 bars.

derek

Sorting shouldn't really have a place in pie charts anyway. Grouping, yes, and sorting within group; but if you think you can line all the groups in a pie chart up in order, then you haven't really got a set of groups suitable for the circular nature of a pie chart, and should go for bars instead.

I guess the exception would be a rock-scissors-paper series, where there is no clear start or end, yet the elements are in a clear order with respect to their neighbors. I can't think of a real-world example just now, though.

david

Like a.c. I fail to see how the JunkCharts version passes the self-sufficiency test; it doesn't really convey much information to me at all, really.

derek

I think the attempt to show everything as a proportion of 100% makes the dark bars too small, so I'd abandon that. Then the rectangles are too slim amd widely spaced for the dark to be what Bertin called "retinal variables". I'd close up all the visually-active white spaces and just have gray gridlines, and re-order the categories slightly, a bit like this.

Kaiser

David: A few things are apparent from the chart, such as Carbon dioxide overwhelms pretty much everything else, that main CO2 emitting sources do not emit other types of gases,that methane and nitrogen oxide are afterthoughts (at least in terms of volume).


Derek: your version looks more attractive. what is the scale you're using? Which part adds up to 100%? I agree with you that the chart can be improved by removing all white space in the interior.

derek

Jon, the scale there is about 40% from left to right. I tried 100%, with the baselines at 9% and 9+18%, giving the three components their respective spaces in 9%, 18% and 72%, but as I said about Kaiser's bars, the 100% space in a an unstacked bar of eight components created a vast area (800%) of space, that I felt lacked value.
This way, all the dark areas add up to 100%, and the white areas are 8x40%-100%=220%, enough for space, but of no significance to the total.

But I don't feel 8x100%=800% of space has significance either. Most people wouldn't do that with a regular bar chart, even if it was a bar chart whose components added up to 100%.

derek

Not Jon, Kaiser, sorry!

LoveStats

Edward Tufte would be proud of this ridiculous chart. It can certainly be nominated to the hall of fame!

The comments to this entry are closed.

BOOTCAMP SUMMER '19



Link to Principal Analytics Prep

See our curriculum, instructors. Apply.
Kaiser Fung. Business analytics and data visualization expert. Author and Speaker.
Visit my website. Follow my Twitter. See my articles at Daily Beast, 538, HBR.

See my Youtube and Flickr.

Book Blog



Link to junkcharts

Graphics design by Amanda Lee

The Read



Keep in Touch

follow me on Twitter

Residues