retire data process functions in chevron

insightsengineering / chevron

Standard TLGs For Clinical Trials Reporting

https://insightsengineering.github.io/chevron/

Other

11 stars 1 forks source link

retire data process functions in chevron #154

Closed wangh107 closed 2 years ago

wangh107 commented 2 years ago

retire functions that are migrated to dunlin @BFalquet please help update the following mapping and add missing items if not listed below.

[ ] chevron:::pivot_wider_labels ---> dunlin::poly_pivot_wider
[ ] chevron:::h_as_factor ---> dunlin::h_as_factor
[ ] chevron:::reorder_levels_params ---> dunlin::co_relevels
[ ] chevron:::dm_explicit_na. ---> dunlin::dm_explicit_na
[ ] chevron::cut_by_group ---> dunlin::cut_by_group The functions are used in chevron.sample.study, create a github issue there to change dependency

BFalquet commented 2 years ago

@wangh107 After careful consideration, I am of the opinion to keep var_labels_for in chevron. This function is not used in preprocessing (for now) and is deeply integrated in many chevron TLG functions. In addition, it depends on tern. Moving it to dunlin would make it depend on tern for very small benefits.

BFalquet commented 2 years ago

In addition of removing the function definition, please:

[x] Add downstream dependency to chevron in dunlin.
[x] Create a branch with the same name in chevron repo and add dunlin as an upstream dependency.
[x] Replace pivot_wider_labels by poly_pivot_wider in the body of the syn_test_data function of chevron. (be mindful that the result is now a list, adapt the test in consequence)
[x] Replace reorder_levels_params by co_relevels in the body of the reorder_adex_params function of chevron.

Cheers