Foundation
s olor for Data
C
Vis
Benjamin
Bach
http://benjbach.me
University of
Edinburgh
202
0
Outline
What is color?
Color in
visualization Color
Scales
Sequential, diverging, categorical
scales Rainbow color map
Color blindness
What is
Color?
~120 million
400-700nm rods
~6 million cones
Color Models
Subtractiv
e
Models
What is
color?
HSB Model: Hue, Saturation,
Brightness
Color = (Brightness/Lightness, Saturation,
Hue)
Color in
Visualization
Visual Variables
http://innovis.cpsc.ucalgary.ca/innovis/uploads/Publications/Publications/Carpendale_visual_v
ariables.pdf
Color
Scales
More examples of color scales
(ColorBrewer)
Colorbrewer.org
Sequential
Scales
More examples of color scales
(ColorBrewer)
Sequential color
scales
Single
hue:
Multi
hue:
https://github.com/d3/d3-scale-chro
Mapping color
scales
Data Linear
value
Non-linear
Stepwise
Median
Median
Median
Median
10
8 1 %
8
10
%
Diverging
Scales
More examples of color scales
(ColorBrewer)
Diverging color
scale
Multi
hue:
Categorical
Scales
More examples of color scales
(ColorBrewer)
I Want
Hue
Rainbow Color
Maps
More examples of color scales
(ColorBrewer)
http://innovis.cpsc.ucalgary.ca/innovis/uploads/Publications/Publications/Carpendale_visual_v
ariables.pdf
Color Maps
Rainbow
:
Diverging
:
Rainbow Colormap
Sequential/Divergent Colormap
Color (mis)use
Lightness perception of rainbow color
maps
Rainbow colormap
Pros:
- Many different values
- Compare similar values
Cons:
- Colors hardly orderable
- Introduces sharp jumps in values
- Hides overall patterns
- Overemphasizes certain values
- Is not black/white nor colorblind
save!
Linear
Rainbow
map
Color
Blindness
Normal vison
Deuteranopia
Protanopia
Tritanopia
Protanopia Deuteranopia Tritanopia
"If a submitted manuscript
8% happens to g o to three
male reviewers of Northern
European descent, the
0.5% chance that at least one will
be color blind is 22
percent."
www.color-blindness.com
Normal
Types of color
blindness
Colorblind safe
palettes
Colorbrewer.org:
Diverging
Sequential Categorical
Redundant encoding
Further applications in
DataVis
Meaning of colors across
cultures
Take home messages:
● Color = hue + lightness + saturation
● Chose a color scale that fits your data
● Avoid rainbow color palettes
● Test your color palettes for colorblindness
● Whatever increases accessibility for some
fraction of the population, increases
accessibilty overall.
Further reading
● Wong, Bang. "Points of view: Color blindness." (2011): 441.
● Tamara Munzner: Visualization Analysis and Design,
2016; Chapter 10: Map Color and Other Channels
● https://www.color-blindness.com/types-of-color-blindnes
s
● Colin Ware: Information Visualization—Perception for
Design, 2012
