We are accustomed to seeing colour as one "channel", but as Munzner mentions in Chapter 5 of Visualization Analysis and Design, we can actually split color into at most three useful channels.
One way in which I can imagine colour channels to be split is into hue and saturation. For example, if I wanted to encode the time of transactions in HDB resale prices by color, I could add two useful pieces of information: the hue (e.g. yellow to red) being the average month of all transactions, and the saturation being the range (e.g. standard deviation) of the transactions.
This way I can tell, if I found the time of transactions to be important, that certain groups of data have less reliable averages because they have a wide range.
Implementing something like stroke-hue and stroke-saturation would be a great way to describe dataset distribution.
We are accustomed to seeing colour as one "channel", but as Munzner mentions in Chapter 5 of Visualization Analysis and Design, we can actually split color into at most three useful channels.
One way in which I can imagine colour channels to be split is into hue and saturation. For example, if I wanted to encode the time of transactions in HDB resale prices by color, I could add two useful pieces of information: the hue (e.g. yellow to red) being the average month of all transactions, and the saturation being the range (e.g. standard deviation) of the transactions.
This way I can tell, if I found the time of transactions to be important, that certain groups of data have less reliable averages because they have a wide range.
Implementing something like
stroke-hueandstroke-saturationwould be a great way to describe dataset distribution.