ejikeugba / serp Goto Github PK

I see that you are using cat() whenever you want to communicate with the user. It is considered better practice to use message() instead so that it is directed to the correct output (stderr instead of stdout) and it can be silenced by the user via suppressMessages().

See this great stack overflow answer for a longer explanation: https://stackoverflow.com/a/36700294/4439357

This issue is part of openjournals/joss-reviews#3705.

Here are some suggestions to improve the manuscript.

1. Reference section:

• To order the references in the same order that they are cited in the manuscript.
• Line 119, the url of the DOI has an error: https://doi.org/doi.org/10.4137/CIN.S20806

2. Table 1 and Figure can be made to be more self explanatory, there are abbreviations (, vglm, NPOM, lambda, ..) that are explained in the main text only. Please can you add a footer for the aberrations to table 1? The same also applies to the figure 1, you can give the full names.
3. Should line 64 be A Minimal Example ?, “m” is lower case.
4. Can the section called “Statement of Need” be renamed “State of the Field” ? Since this section describe how this software compares to other commonly-used packages.
5. You can add another section below the introduction called “Statement of need” that clearly states what problems the software is designed to solve and who are the target audience. Part of the content is already mentioned in the second paragraph of the summary section.

openjournals/joss-reviews#3705

As noted in the whedon check, you need DOIs for a few of the refs.

MISSING DOIs

10.2307/2288515 may be a valid DOI for title: Advanced Econometrics: A Bridge to the Literature
10.1007/bf02985802 may be a valid DOI for title: The Elements of Statistical Learning: Data Mining, Inference and Prediction
10.18637/jss.v099.i06 may be a valid DOI for title: Regularized ordinal regression and the ordinalNet R package

openjournals/joss-reviews#3705

I see your package has a good code coverage but upon closer inspection of what is tested, most of the tests related to errors. It would be good to have some regression tests, where you compare the numerical output of some functions to what is known to be correct (e.g., from earlier versions). This prevents the introduction of a bug that would return incorrect values in the future.

Snapshot testing in testthat provides a convenient way to address this issue: https://testthat.r-lib.org/reference/expect_snapshot.html

This issue is part of openjournals/joss-reviews#3705.

https://docs.github.com/en/communities is a good guide to follow. Ends up being added in the markdown files for the repo, and then also shows up as part of the GitHub pages pkgdown site.

openjournals/joss-reviews#3705

I didn't notice this package had a pkgdown website until it was mentioned in openjournals/joss-reviews#3705. I think it would be helpful to add here in the usual places where people look:

• in the DESCRIPTION file, alongside the GitHub repository. You can see the example of some popular packages such as ggplot2 or data.table
• in the GitHub repository details as described in this stackoverflow answer

Additionally, you should add it in your _pkgdown.yml file to enable cross-linking from other pkgdown websites, as explained in the downlit README.

@ejikeugba - I suggest editing the README.md and add a short section at the top on dependencies. Yes, CRAN handles it, but it's good to warn the user of incoming forced-installs of packages.

openjournals/joss-reviews#3705