this post was submitted on 20 Oct 2024
85 points (84.6% liked)

Data is Beautiful

1114 readers
36 users here now

Be respectful

founded 4 months ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 55 points 2 weeks ago (1 children)

Line graphs of percentages not based at zero make it difficult for me to grok the magnitude of changes. Missed opportunity for the hue line color to match the actual hue in the vertical. Just being an angle value I have no idea what hue it's supposed to be.

[–] [email protected] 10 points 2 weeks ago (1 children)
[–] [email protected] 2 points 2 weeks ago (1 children)

Not a colorologist, but how the hell do you measure saturation and hue in a poster?

[–] [email protected] 4 points 2 weeks ago (1 children)

Not a colorololologist either, but I'm guessing, you scan it and then calculate the average of the individual pixels. You can convert RGB color values to HSV color values to get the hue, saturation and value.

But calculating the average of the hue degrees (like was done for this post) is non-sense.
0° is red, as is 360°, so you get completely different averages, depending on how you calculate them:

  • (0°+120°+240°) / 3 → 120° (green)
  • (360°+120°+240°) / 3 → 240° (blue)
[–] [email protected] 2 points 1 week ago (1 children)

Does hue make sense at all? If yes, you could calculate the vector mean (convert degrees to x and y axis, average, and convert back to degrees) and get something sensible out of that.

[–] [email protected] 1 points 1 week ago

Ah yeah, good idea.

But good question, if it makes sense at all. Since hue is not a linear scale, the average is not exactly the 'middle' of anything. At best you could interpret it as the poster being somewhat biased in that direction, but ultimately you could have a poster with no green at all, which still averages out to green, and that just doesn't feel right.

A more complex graph, which shows the frequency of the colours used, might be more useful...