Use TidierPlots for some of the graphics

[dmbates]

It seems that AlgebraOfGraphics.jl is not being actively developed and I decided to take a look at TidierPlots.jl as a possible replacement. At present some of the facilities for comparative plots still need refinement. Faceting is broken after the switch to Makie as a back-end and vector color/colour arguments in aes are often interpreted as if they were color specs instead of categorical vectors that determine the color of the geom.

This branch provides a few plots with TidierPlots.jl for evaluation of whether this is a good path to follow.

[palday]

I think this is looking good.

AoG is getting some love again (https://github.com/MakieOrg/AlgebraOfGraphics.jl/releases/tag/v0.7.0) but switching over to a plotting package that's closer to everybody's favorite R package is probably better for reaching a broader audience. (=I'm still onboard with swapping to TidierPlots for books and teaching.)

[palday]

Suggested change

	# draw(
	# data(gausshermitenorm(9)) *
	# mapping(:abscissae => "Positions", :weights);
	# figure=(; size=(600,450)),
	# )

[palday]

Suggested change

	# draw(
	# data(gausshermitenorm(9)) * mapping(
	# :abscissae => "Positions",
	# :weights => log2 => "log₂(weight)",
	# );
	# figure=(; size=(600,450)),
	# )

[palday]

Suggested change

	# draw(
	# data(@subset(dsm01pars, :type == "σ")) *
	# mapping(
	# :value => "Bootstrap samples of σ";
	# color=(:group => "Group"),
	# ) *
	# AlgebraOfGraphics.density();
	# figure=(; size=(600, 340)),
	# )

[palday]

Suggested change

	# #| warning: false
	# let wlen = 1:21
	# draw(
	# data((; wrdlen=wlen, count=counts(byitem.wrdlen, wlen))) *
	# mapping(:wrdlen => "Length of word", :count) *
	# visual(BarPlot);
	# figure=(; size=(600, 450))
	# )
	# end

[palday]

Suggested change

	# draw(
	# data(pruned) *
	# mapping(:rt => "Response time (ms.) for correct responses") *
	# AlgebraOfGraphics.density();
	# figure=(; size=(600, 340)),
	# )

[palday]

Suggested change

	# draw(
	# data(pruned) *
	# mapping(
	# :rt =>
	# (
	# x -> 1000 / x
	# ) => "Response speed (s⁻¹) for correct responses",
	# ) *
	# AlgebraOfGraphics.density();
	# figure=(; size=(600, 340)),
	# )

[palday]

Suggested change

	# draw(
	# data(@subset(bxm03pars, :type == "ρ")) *
	# mapping(
	# :value => "Bootstrap replicates of correlation estimates";
	# color=(:names => "Variables"),
	# ) *
	# AlgebraOfGraphics.density();
	# figure=(; size=(600, 400)),
	# )

[palday]

Suggested change

	# draw(
	# data(@subset(pnm01pars, :type == "σ")) *
	# mapping(
	# :value => "Bootstrap samples of σ";
	# color=(:group => "Group"),
	# ) *
	# AlgebraOfGraphics.density();
	# figure=(; size=(600, 340)),
	# )

[palday]

Suggested change

	# draw(
	# data(@subset(psm01pars, :type == "σ")) *
	# mapping(
	# :value => "Bootstrap samples of σ";
	# color=(:group => "Group"),
	# ) *
	# AlgebraOfGraphics.density();
	# figure=(; size=(600, 340)),
	# )

[dmbates]

It appears that an earlier issue, TidierOrg/TidierPlots.jl#102, has resurfaced. I think we should delay merging this PR until that issue is resolved. It is probably also related to the mismatch of symbol color and line color is some of the plots in largescaleobserved.qmd.

[ajinkya-k]

Can this be merged now?

[palday]

@dmbates -- over in #80 @ajinkya-k has proposed converting everything back to AoG, now that Julius has taken up maintenance and development of AoG.

I don't have a strong preference for which system to use, but we should probably pick one and use it consequently. I'm more familiar with AoG, but of course TidierPlots has the explicit goal of being more familiar to R converts. What are your thoughts?