Points and lm, how to ensure nothing gets clipped

[kylebutts]

Hi Grant, is there a way to ensure that points and/or the linear model don't get clipped outside the axes? One idea would be to pass a character vector to type to ensure this?

library(tinyplot)
tinyplot(Sepal.Width ~ Petal.Width | Species, data = iris, type = "lm")
tinyplot_add(type = "p")

tinyplot(Sepal.Width ~ Petal.Width | Species, data = iris, type = "p")

^{Created on 2025-08-21 with reprex v2.1.1}

[grantmcdermott]

Hi Kyle, I'm afraid you've hit upon the first tinyplot gotcha that I highlighted in my useR! presentation: https://grantmcdermott.com/tinyplot/vignettes/useR2025/useR2025.html#/layering-gotchas

I don't really have a good solution ATM besides switching the order:

tinyplot(Sepal.Width ~ Petal.Width | Species, data = iris, type = "p")
tinyplot_add(type = 'lm')

(which works in a lot of settings, but can also end up clipping in reverse.)

Or, setting x/ylim manually in your call:

tinyplot(Sepal.Width ~ Petal.Width | Species, data = iris, type = "lm", ylim = c(2, 4.5))
tinyplot_add(type = 'p')

One idea would be to pass a character vector to type to ensure this?

Yeah, this was actually something we discussed way back in #93. (I make the point about different x/y ranges obliquely in the opening post.) Unfortunately, supporting this kind of functionality now will require a lot of work due to the way our type processing works internally; i.e. we construct a dedicated datapoints object for that specific type, which is then used to determine the plot range etc. It might be possible to wrap multiple types in a meta lapply() call and then take it from there (e.g. so we can keep track of multiple datapoints objects internally). But I don't have the time to look into that rn, unfortunately. Sorry!