crsl4 / fast-stats Goto Github PK

Overall this looks like a handy contribution. The apps provide a user-friendly way to interact with data for a fairly defined project.

I have a few suggestions below.

Authorship

• Yizhou Liu seems to have contributed one of the slide decks and a fair bit of the commits. I wonder if they should be included in the authorship? Or acknowledged more in the paper?

Functionality & Documentation

• As noted in #52 , thatssorandom couldn't be installed. I was able to get it working with that issue's guidance.
• Overall, I think this section needs work. Starting the application as indicated does not work.
• When running the webinar-aug20 app to make a MosaicPlot, I got the error:

  Warning: `guides(<scale> = FALSE)` is deprecated. Please use `guides(<scale> = "none")` instead.

• I think expanding documentation would help facilitate using this tool in teaching. For example, the error messages for making plots can be a bit frustrating--maybe identify which columns of the dataset are categorical vs numerical? Or provide more description of the sample dataset? Or limit selections to the appropriate variable type?
• I like that the slides are part of the github repo, would it be possible to share the Google Slides link as well?
  • Would it be possible to share the datafiles used in making the figures in the slides? Or provide a copy that matches the sample file? In going through the slides, I tried to re-create the graphs and while similar, that are different. I assume different data were used, but would be nice to be able to replicate the examples provided.
  • I think this could also serve as the start of adding tests. I didn't investigate the code for the Chi-square, etc, Should probably add tests.
• Warning: Groups with fewer than two data points have been dropped. is reported in R for some of the plots, but no feedback is given to the App. Maybe that should be passed through?

notebooks folder

Not really sure what this folder is supposed to be. The notebooks do not work as-is. e.g. there is no data/ecosystem-data.csv file to load as on line 18 of ecosystem-data-analysis.Rmd.

Software paper

• References: Please add DOIs to all papers.

Editorial

• DOCS.md: "The Home page is a collection of animation using WI Fast Plants photos.": "collection of animation" seems like a poor word choice here.
• Webinar links in timeline seem broken. Not necessarily something that needs fixing, but might want to either fix or remove links.
  • https://www.carolina.com/knowledge/2020/10/20/webinar-student-designed-ecosystem-experiments-in-middle-grades - 404 Error
  • https://fastplants.org/2020/08/06/new-fast-plants-polycots-selection/ - Redirect to main site

Hello!

First, great job on the apps, they look great and I am sure they will indeed be a fundamental step in teaching data science. I found the documentation thorough and clear, and the apps engaging and fun to use. Please find below my comments associated with the checklist in the JOSS review. Please let me know if you have any questions!

General checks

Everything looks good!

Functionality & Documentation

I was able to download the repo successfully using the provided command. However, I had several issues with the rest of the provided code.

• The package thatssorandom is not available on at least the two most recent versions of R, so I had to use install.github(“https://github.com/EdwinTh/thatssorandom") to get it and use the apps
• I could not run the app using the provided command (shiny::runApp("/path/to/wi-fast-stats")). I think the first issue here is that the repo downloads as fast-stats and not wi-fast-stats. Then, the provided command should directly point to the folder where one of the app is present in shiny-app/.
• I think the code could use a bit more commenting. It took me some time to wrap my head around what was happening. More importantly, as this app is intended to a younger audience, it could be their entry door to coding. If the code is adequately commented, it will definitely be more understandable by students (and teachers/researchers!) and thus have broader impact.

In terms of the app functionnality, I only have two main comments (only using sample data):

• I think use of the app would be made a lot easier if the "Group variable" and "Quantity" radiobuttons could only show the corresponding variables, instead of all the columns (categorical and numerical variables, respectively).
• It may be worth it to let plant ID be a "Group variable"? I am not sure if yes, but my data mind would love to see a raw data plot or a histogram.

Software paper

• Williams and Hill (1986) is missing DOI: 10.1126/science.232.4756.1385
• The paper refers to a section called "Data summary" but the apps have either nothing or "Data upload", I would make section names constistent.

Pierre

See the comments on commit 1 and commit 2