r - appropriate way to save/load trained models in ShinyApps deployed with docker

Question

I trained a xgboost model with tidymodels:

model = boost_tree() %>%
  set_mode("classification") %>%
  set_engine("xgboost") %>% fit(variable ~ ., data)

and saved it as follows into my app directory:

saveRDS(model, "model.RDS")

In my Shinyapp code i read it with:

readRDS("model.RDS", .GlobalEnv)

This worked fine until i yesterday updated rstudio and r (from 3.6.2 to 4.0.3) to newest versions on my local desktop. It also works fine when i run the ShinyApp on my local Desktop with the newest versions. However, when i deploy the ShinyApp to my Server with Docker/Dokku i get the error message:

Warning in gzfile(file, "rb") :
  cannot open compressed file 'model_m2.RDS', probable reason 'No such file or directory'
Error in gzfile(file, "rb") : cannot open the connection
Calls: <Anonymous> ... source -> withVisible -> eval -> eval -> readRDS -> gzfile
Execution halted

I build my Dockerfile from A rocker baseimage rocker/shiny-verse:latest.

Does anyone have an explanation for this behavior? I can provide more information about my setup if needed. Thank in advance

question from:https://stackoverflow.com/questions/65904436/appropriate-way-to-save-load-trained-models-in-shinyapps-deployed-with-docker

Answer 1

Waitting for answers