Why is GGally::ggpairs significantly slower in RStudio vs. base R?

Question

Per the title, does anyone know why rendering a ggpairs plot from the GGally package takes significantly longer in RStudio vs. base R (or terminal)?

Example:

start.time <- Sys.time()
ggpairs(mtcars)
end.time <- Sys.time()
time.taken <- end.time - start.time
time.taken

Running this in RStudio on my machine takes on the order of 5 times longer than base R. I have experienced the same slow down regardless of OS (Windows vs. Mac).

Are there any workarounds?

Other packages?

Specifically, how to render something like: GGally::ggpairs(iris, color = "Species") quickly without leaving RStudio?

Answer 1

I had similar issues, and spent some time trying to figure out why. I found four significant issues (not an exhaustive list). If your situation is like mine, then 1 and 2 are your main concerns.

1. The IDE. RStudio performs ggpairs slower than R.
2. Your computing environment. I don't have the resources to test this extensively, but we're most likely talking about GPU, since this is a graphics processing issue.
3. Number of variables. More variables = exponential growth in time.
4. Sequential operations. If you are working on a low-power machine, you could slow execution with too many requests.

You can read more on my github here: https://github.com/zstachniak/Elapsed-Time-Pairwise-Functions/blob/master/ggpairs.md