Skip to content

Consider deprecating some ways of specifying input computation in epi[x]_slide #629

Open
@brookslogan

Description

@brookslogan

The slide function documentation is wordy and complicated, and the code is very complicated.

Consider making epi_slide() support only the tidyeval case, removing .f and .new_col_name. Maybe consider moving .window_size before ... so one could do edf %>% epi_slide(28, md_rate = median(rate)). Would play better with tibble(outcol1 = ...., outcol2 = ....) tidyeval feature, as there is no need to add an extra comma to make .f missing. Would also play better if we eventually add across support. Potential drawbacks:

  • Tidyeval overhead may mean this is significantly slower than alternatives.
  • Requires some arcane syntax for more complicated computations; my_growth_rate = { <multiple statements> }.

In epix_slide(), function and formula .f and outputting data frames, not ordinary vectors, is more common, though data-masking and ordinary-vector output might be used for some max time_value / reporting latency calculations. We could potentially force function/formula usage.

It'd also be good to do this at the same time as, or after, renaming epi_slide() and epix_slide() to more distinct names.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions