turn visualizations into geoms

[bvreede]

We want our conversation visualization to be geometries that can be included in a ggplot.

Some useful material:

• a tutorial on making a geom from Rladies Melbourne
• A gallery full of inspiration with other ggplot-compatible libraries

See also #14

[n400peanuts]

@n400peanuts

[mdingemanse]

In my experience with ggplot and the tidyverse, the best geoms are conceptually simple and have a clear use case. The two first and conceptually simplest examples that come to mind for the kind of data we have in {talkr} are these:

1. geom_turn(), initially described in add geom_turn() functionality #14
2. geom_wave() (Add geom_wave #46 ) — the latter only possible if there's audio data on hand

I find it hard to think of other equally basic geoms at this point. Given these two geoms (or even only the first), all kinds of manipulations and plot styles come within reach through the logic of ggplot — for instance:

• changing coordinate systems as here
• mapping single conversations vs sets of extracts as illustrated in separate sampling multiple (uid-focused) from plotting 1 conversation (source-focused) #18
• doing a vertical version as in vertical visualisation: is this something? #11