Internal identification provides information about the choice of encoding tools.
Internal identification is particularly useful when encoding complex and multidimensional data and can be achieved in one of two ways: a legend or on-the-plot identification.
A legend describes the variation in encoding choices that reflects variation in the data, in a separate area usually outside the data region.
In the graph example discussed in graph identification, the legend identifies the point visual implantations as the data, and the line visual implantations as a linear predictive model on the data. The legend also identifies the blue colour points and line from the group ‘Class A’, and the orange colour points and line from the group ‘Class B’ (two example qualitative groups in the data).
The alternative to legend is on-the-plot identification.
Best practice suggests that where possible it is always better to internally identify encoding choices directly on-the-plot region rather than placing a legend outside in the graph region.
As an example, consider a line chart that shows the sum of goals scored by every soccer team participating in the Italian 2018-2019 Serie A season. There were 20 teams competing, and a legend approach would require encoding 20 different colours one for each line, and the then identifying each color with the respective name of the team in a legend, as for instance:
This is a messy approach and decoding is simply impossible using a legend. We cannot identify the colours associated with each team, and it takes great mental effort to go back and fort from the legend to the line chart. Also notice how the legend takes about 1/3 of the graph’s area, and this is not data.
The solution to this problem is on-the-plot internal identification. This frees much space for graphing the data, and importantly it eliminates ambiguity and enables instantaneous decoding.
Notice also how it is unnecessary now to use many colours, and all you have to do is alternate
Notice also how it is unnecessary now to use many colours, and all you have to do is alternate between two or three colours. Here, I choose three colours that are colour blind safe, as sourced from Color Brewer. I also encode Juventus in black as the team that won the championship (Juventus’s colours are black and white).