tweaks to figures, moved tidierplots figures to AoG

[ajinkya-k]

longitudinal.qmd

• all figures now in pure julia
• cleaned up some legends
• fixed a typo (Itercept) -> (Intercept)

largescaleobserved

• moved all figures to AlgebraOfGraphics
• added extra panel to L22 vs full memory size plot
• this allows removing TidierPlots as a dependency, which will allow us to set alpha for histograms in intro.qmd (see below)
• this allows us to remove RCall as a dependency (which should get rid of Conda)

intro.qmd:

• added alpha to histograms

[ajinkya-k]

preview here: https://ajinkya-k.github.io/EmbraceUncertainty/longitudinal.html#the-elstongrizzle-data

[ajinkya-k]

There is just one more plot left to move from R to Julia

[ajinkya-k]

@palday can we get this merged?

[ajinkya-k]

this will get rid of most if not all R dependency

[palday]

ideally we should be able to set this at the project level and not need to specify it in every file

[noblock]

[palday]

a little bit of history here: we were actually working on moving in the other direction, i.e. porting all the AoG plots to Tidier. There were two reasons for this:

• AoG had become unmaintained (no longer an issue since Julius reinvigorated the project)
• @dmbates thought that the Tidier ecosystem might be more accessible to new/potential converts from R

I don't have a terribly strong preference either way, but I don't want us to be bouncing back and forth as TiderPlots and AoG one-up each other.

I think porting R-based figures to Julia is an obvious improvement -- could you split that out into a separate PR that I'll review and then we can leave the bigger Tidier vs AoG discussion for this PR?

Also, I haven't checked the render yet, but make sure the lattice plot in Julia matches the aspect ratio in R. @dmbates has noted that the human eye is very good at pickup deviations from a 45 degree diagonal and so it's good to have an aspect ratio that makes that diagonal meaningful. (I'm doing a horrible job explaining this.)

[ajinkya-k]

@palday see #92 which only has R -> julia plot conversions

[ajinkya-k]

@palday I dont understand the BADCERT_EXPIRED error i get when trying to preview my PR on my fork. See here. This error does not occur on my local

Any idea what this is?

[palday]

Your local copy has that data cached, so it doesn't have to re-download it, but the CI runner needs to pull it down. It looks like the certificate expired at Tue, 13 May 2025 19:02:04 GMT, so that's why it just now started failing.

[ajinkya-k]

Your local copy has that data cached, so it doesn't have to re-download it, but the CI runner needs to pull it down. It looks like the certificate expired at Tue, 13 May 2025 19:02:04 GMT, so that's why it just now started failing.

I see. any idea how i can fix it?

[palday]

Just have to wait for the site owner to fix their certificate. Hopefully they notice soon...