prediction-based plots when using by matrices in linear functionals about gratia HOT 3 CLOSED

Thanks for the heads-up @nicholasjclark; the handling of mgcv's linear functional capabilities is pretty limited in {gratia} just now. Some of the model forms allowed by mgcv are supported (not that this is documented anywhere though); DLNM type formulations work, but I don't think I've tried any of the models with matrix by terms.

@dill has mentioned that he and a colleague have been thinking about better displays for linear functional terms.

Once I get v0.9.0 shipped to CRAN I can take a closer look at these things. But I'll look at this specific example before hand to see if it's an easy fix.

from gratia.

I spoke too soon; seems this has already been fixed in the GH version. With

draw(mod)

I now see

which seems to correspond to what mgcv:::plot.gam() produced.

from gratia.

Perfect thanks @gavinsimpson, that update has fixed my issues. I've used these in my blog post on group-varying distributed lag terms, glad I could now include {gratia}.

@dill, I'm looking forward to what you come up with in terms of visualising and communiticating these effects. I always find them challenging to dissect

from gratia.