[OC] Ein Spinnendiagramm ist ein ansprechendes Bild, aber es verdeckt Informationen. Meiner Meinung nach bietet ein Balkendiagramm mehr Kontext, integriert beide Datensätze und ermöglicht es dem interessierten Betrachter schließlich, schnell die (relativ) übergroße Trennungsleistung der besten Viking-Modelle zu beobachten.

Original chart and data taken from https://x.com/LombardiHimself/status/1835464839017140522. The bar charts were drawn by chatgpt using matplotlib. If there’s enough interest, I’ll add to my repo.

Curious what your perspectives are on this.

I think the plot choice isn’t really the issue here, your 2 issues with the spider charts seem to be the mismatched scales and an average benchmark that doesn’t jump off the page. With a scale fix and a visual shift, those both become non-issues.

The bigger problem to me is the perceived “area” of the chart. If you reorder the receivers, the area will be different.

So to fix this, 1) sort the receivers clockwise for most to least 2) line up the scales 3) make the indication for the “average” ring stand out 4) clarify if the average line is only using starters or all receivers, since that changes context a lot