This PR refactors the code into a targets pipeline. It also adds a DESCRIPTION file to manage dependencies.
For the simulation results, I switched to an arrow dataset. arrow is great for large datasets that need to be filtered efficiently. I gitignored the data/ folder where the arrow dataset lives. You can just use dplyr to work with the arrow object, followed by collect() to bring the data into R. Only the data_path target is an arrow dataset; model_results is a normal tibble.
Closes #3
This PR refactors the code into a targets pipeline. It also adds a
DESCRIPTIONfile to manage dependencies.For the simulation results, I switched to an arrow dataset. arrow is great for large datasets that need to be filtered efficiently. I gitignored the
data/folder where the arrow dataset lives. You can just use dplyr to work with the arrow object, followed bycollect()to bring the data into R. Only thedata_pathtarget is an arrow dataset;model_resultsis a normal tibble.Add new targets to
tar_plan()in_targets.R. New targets should be generated by functions located inR/.Run
targets::tar_make()to build the pipeline. You can also use the build pane in RStudio or the build shortcut (cmd/ctrl+shift+B)Results are cached in
_targets.