Next steps?

[nestanyol]

@lwjohnst86 Luke, which steps are we missing for building the R package? :) Any recommendations?

[lwjohnst86]

Yes! I plan on looking over what you've done now over the weekend and next week and make some issues, do some reviews and stuff 🤩

[lwjohnst86]

• Many of the test_ files contain functions that are very very similar. Maybe try to consider ways to simplify it so you don't need to repeat the same code throughout them all.
• Many functions output a list of dataframes. It might be a good idea to think how your functions can be created so that only a single dataframe is output. Trying to follow the principle of "Object type as input, same object type as output" (https://design.tidyverse.org/out-type-stability.html), particularly the "tibble in, tibble out" principle makes things easier to chain together with the pipes. Likewise, if the function takes a list, then the best practice is to return a list.
• Many of the functions need more descriptive Roxygen titles (e.g. test_knn_regression() has the title test_model_knn_regression, which isn't very descriptive.
• While technically return() is used to output the end result of a function, in practice, we don't use it. See more details here: https://style.tidyverse.org/functions.html#return
• I did some PR's with comments, and also accidentally pushed to the main branch, but I have some commits there with some comments as well.
• Consider naming your functions as verbs (e.g. instead of preprocess_data(), call it simply preprocess()).
• Some of your arguments' names are shortened cutoff_subj, consider keeping the full word instead for clarity of description.
• It's time to start writing some unit tests! See #37 for some details about that! To run unit testing while in RStudio, use Ctrl-Shift-T.
• I've added some infrastructure files to have your code get tested and checked on GitHub, automatically. See #38
• You'll eventually need to create your documentation using devtools::document() (Ctrl-Shift-D). To run that, you need to merge #39 first. You'll probably encounter a lot of errors and warnings as you update/fix your documentation within the package.
• Once you've finished with most of these, you can start running local CRAN checks for basic checks of your package. You can do that by running Ctrl-Shift-E. But only do that later, after dealing with all of theses :grin:

[danielaags]

@lwjohnst86 Nuria told me that it could be a good idea to join the training and testing functions into a single one. Shall we also put that in the to do list?

[lwjohnst86]

In general, it's best to keep functions small and very focused/targeted. So keeping the testing and training separate is better than having them together, especially since, looking at their code, they have quite a bit different in them.

[danielaags]

I have implemented the suggestions in all the functions. I would like to start working on the unit tests, how shall I move forward? Shall I merge the unit test branch into main and the pull it in my branch (DDLS branch)?

Update: I followed the steps described in the test branch, got the folder and started working on it, I have unit test for all the functions now. It helped me a lot the example from Luke!