check_model for multiple models?

[candour23]

Hi,

This is my first brave question on github. First of all - great package- thanks a lot!

Second, I am running multiple linear regressions, but my knowledge is vey limited in programming and purr.
I found a helpful site that helped me run multiple regression models (different outcomes) here: https://aosmith.rbind.io/2019/07/22/automate-model-fitting-with-loops/

But then, I wanted to use your beautiful package to check my assumptions on many models (different outcomes)... and I wasn't able to pull it off.
I would need to fit one by one model, and plot one by one (which is great until you have >3 models).

My ideal scenario would be having models in table as the link above explains (which I have),
and then using your fancy package to plot all these models in a single pdf, with names of output (for each separate model) corresponding to each of model, so I could easily detect something off.

I know this is going a bit beyond the scope, but if you have some suggestions, would be appreciated.
Or if you know an easy way to integrate multiple models in check_model + name of each different outcome?
Kind of just continuing on the first link I provided and relating it with check_model function?
Thanks in any case, great package!

[bwiernik]

So once you've got your list of model objects model_list --lapply(model_list, check_model) or map(model_list, check_model) will apply the check_model function to each model in the list.

[bwiernik]

assign that result as an object plot_list <- map(...) and then extract the results by name plot_list$mpg

[candour23]

I am not sure if I am not grasping your concept, or if you are on the same page with me for the question.
For each model I will have 5 plots (just like first solution you gave), what would be useful is to above these (or anywhere) an actual name of the plot (Model 1) above the plot itself, then for next 5 plots- Model 2, etc.
Can this be embedded on the overall plot itself?
So, if I am checking 25 models from single pdf or however I save it, and I see one really odd, from the pdf itself I can say this exact model with this outcome (written above it) is violating my assumptions?

[bwiernik]

At the moment, I suggest using headings in your RMarkdown document to do this, ala:

library(purrr)
library(patchwork)
library(performance)

model_list <- map(
  c(disp = "disp", hp = "hp", drat = "drat", wt = "wt"), 
  ~ lm(as.formula(paste("mpg ~ ", .x)), data = mtcars)
)

plot_list <- map(
  model_list,
  check_model
)

Model Checking

disp

plot_list$disp

hp

plot_list$hp

[bwiernik]

Doing more customization will be easier if see switches to using the patchwork package instead #293 (reply in thread)

[candour23]

Okay, if markdown cant be avoided, then I'll implement it as you suggested.
Thanks a lot for looking into it and giving me a temporary solution.