Feedback: Tidyverse (Afternoon Session)

[nehamoopen]

Please leave your feedback for the tidyverse component (afternoon session) in this issue, thaaaanks!

[StefanoRapisarda]

These were comments made after following the very first course in R, comments are organised by session (corresponding to videos):

• Intro to Tidyverse

  • I think it would be useful spending few words on what is a “package”. According to the level of the workshop I am sure most of the people ignore the programming meaning of the word. Also, explain well the difference between installing, loading, and using. It could be done metaphorically: I buy tools to do a carpentry job, but I do not bring them with me around the house ALL the time (that would make me waste energy), but only when I am ready to work on my project. When needed, then I select what I need from my belt or bag and I use. I will put on my belt (loading) only those tools I need to that specific job, even if I bought (installed) many more;

• Import Data

  • I think that before starting loading the data frame, a full overview of what they are going to do is necessary (ex. “We are now going to use what we learned in the previous section to load a database, making custom selections according to our needs, running some very basic statistics on selected columns and row, and, finally, visualise our data. In order to do that, we will also use functions that they are not native of R, but that belong to the Tidyverse package”). Sometimes this is done by the speaker in person, sometimes it is not, so why not adding it to the recorded material?

[DorienHuijser]

Hello! I haven't gone through the afternoon session materials (and will not manage before the end of this month). I had only noticed so far that the slides link to uu.nl/rdm, that should be https://uu.nl/rdm, because otherwise the link does not work. Sorry, not super useful feedback ;)

[nehamoopen]

Notes from @Mish-JPFD:

• In the penguins dataset (raw) - there are column names with whitespaces, these should be renamed with underscores.
  • If we change the tsv file, we will need to change the code in the subsequent code chunks...
• Exercise 7: typo, we should use the code we wrote in exercises 5 & 6 - not 4 & 5
• Do we need to rethink the ggplot exercises? Because right now, they don't necessarily 'teach' the basics of ggplot - they pretty much just run the code.

[DorienHuijser]

During session notes:

• [x] Reading in the data from the Rmd chunk requires different code compared to importing manually (the file path in the Rmd should be ../data/penguins_raw.tsv, when run from the entire environment it should be data/penguins_raw.tsv. I think this is a something that we can set in the Rmd or in the settings chunk or something. This goes for all reading and writing of files
• [ ] Exercise 5: penguins_subset <- filter(penguins, ???) > when manually importing the dataset, the dataset is not named penguins, but penguins_raw > This is the case for many exercises, needs check!
• [ ] Exercise 5: perhaps add a reflection question on how to check whether the subsetting worked!
• [ ] Exercise 6: add the "And we want to include the data on Culmen measurements" as part of the bulleted list
• [ ] Exercise 6: penguins_subset is overwritten here, should probably just be a new version of the dataset just in case people get stuck here because the selected columns cannot be found while testing
  • So replace penguins_subset <- select(penguins_subset, ???) with for example penguins_subset_select <- select(penguins_subset, ???)
• [ ] There is also a Base R piping operator now: |> which does not rely on the magrittr package. Perhaps that should be mentioned somehow, for people not using tidyverse in the future. Within the tidyverse the %>% and |> are interchangeable (source)
• [ ] Exercise 7:

Use the code you wrote during exercises 4 and 5, and make a workflow that starts with the data penguins, and subsequently applies your filter and select operations.

should become

Use the code you wrote during exercises 5 and 6, and make a workflow that starts with the data penguins_raw, and subsequently applies your filter and select operations.

• [ ] Exercise 7: again, we should not overwrite penguins_subset because it will end up with errors while testing, not good practice: code should become: penguins_subset_piping <- penguins_raw %>% ???
• [ ] Exercise 7 should be REQUIRED and not optional, the whole rest of the Rmd file uses piping
• [ ] Exercise 8: idem: penguins_subset <- penguins_subset %>% mutate(culmen_ratio = ???) > penguins_subset_mutate <- penguins_subset_piping %>% mutate(culmen_ratio = ???)
• [ ] Exercise 9: We should probably explain the difference between long and wide format, perhaps this is already in the video but you know who watched the video anyways
• [ ] Exercise 9: "If you want some extra training in the use of the pipe operator, complete exercise 6 first." > should be exercise 7 instead
• [ ] We should overall give the good example and name code chunks according to what they are doing :)

[nehamoopen]

Exercise 9 combines two functions: rename and pivot_wider, I would perhaps just split them...maybe do the renaming earlier in the script as well

[nehamoopen]

• [ ] Importing data could be done with a simple read_tsv and read_excel, instead of having to go through the import menu -> copying the code -> having to figure out the concept of delimiters + how to specify them in arguments with (for example) /t -> reformatting the date...

• [ ] After a cleaner section on importing data, you could have a section on some basic cleaning/tidying operations which includes formatting dates (currently combined with importing data) and renaming columns (currently combined with pivoting data).

• [ ] After the above two sections are done with simple, clean functions - we can introduce the participants to piping and have them chain all the three together (import, rename, reformat). Then moving onto the rest of the exercises will be easier...

• [ ] The ggplot section could be simplified significantly. The section could start with introducing participants to the different layers involved in ggplot. There's an image of a pyramid which starts with the data as the base and then you add things like aesthetics and facets etc. on top. Thereafter, you could show participants something like the ggplot flipbook: https://evamaerey.github.io/ggplot_flipbook/ggplot_flipbook_xaringan.html and we could structure our exercises to follow this layered way of working (or at least demo it that way).