[REVIEW]: shapr: An R-package for explaining machine learning models with dependence-aware Shapley values

[whedon]

Submitting author: @martinju (Martin Jullum)
Repository: https://github.com/NorskRegnesentral/shapr
Version: v0.1.1
Editor: @terrytangyuan
Reviewer: @frycast, @expectopatronum
Archive: 10.5281/zenodo.3641831

Status

Status badge code:

HTML: <a href="https://joss.theoj.org/papers/f3ea13b8d2ea46c28ad740fd900831c7"><img src="https://joss.theoj.org/papers/f3ea13b8d2ea46c28ad740fd900831c7/status.svg"></a>
Markdown: [![status](https://joss.theoj.org/papers/f3ea13b8d2ea46c28ad740fd900831c7/status.svg)](https://joss.theoj.org/papers/f3ea13b8d2ea46c28ad740fd900831c7)

Reviewers and authors:

Please avoid lengthy details of difficulties in the review thread. Instead, please create a new issue in the target repository and link to those issues (especially acceptance-blockers) by leaving comments in the review thread below. (For completists: if the target issue tracker is also on GitHub, linking the review thread in the issue or vice versa will create corresponding breadcrumb trails in the link target.)

Reviewer instructions & questions

@frycast & @expectopatronum, please carry out your review in this issue by updating the checklist below. If you cannot edit the checklist please:

1. Make sure you're logged in to your GitHub account
2. Be sure to accept the invite at this URL: https://github.com/openjournals/joss-reviews/invitations

The reviewer guidelines are available here: https://joss.readthedocs.io/en/latest/reviewer_guidelines.html. Any questions/concerns please let @terrytangyuan know.

✨ Please try and complete your review in the next two weeks ✨

Review checklist for @frycast

Conflict of interest

• I confirm that I have read the JOSS conflict of interest (COI) policy and that: I have no COIs with reviewing this work or that any perceived COIs have been waived by JOSS for the purpose of this review.

Code of Conduct

• I confirm that I read and will adhere to the JOSS code of conduct.

General checks

• Repository: Is the source code for this software available at the repository url?
• License: Does the repository contain a plain-text LICENSE file with the contents of an OSI approved software license?
• Contribution and authorship: Has the submitting author (@martinju) made major contributions to the software? Does the full list of paper authors seem appropriate and complete?

Functionality

• Installation: Does installation proceed as outlined in the documentation?
• Functionality: Have the functional claims of the software been confirmed?
• Performance: If there are any performance claims of the software, have they been confirmed? (If there are no claims, please check off this item.)

Documentation

• A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
• Installation instructions: Is there a clearly-stated list of dependencies? Ideally these should be handled with an automated package management solution.
• Example usage: Do the authors include examples of how to use the software (ideally to solve real-world analysis problems).
• Functionality documentation: Is the core functionality of the software documented to a satisfactory level (e.g., API method documentation)?
• Automated tests: Are there automated tests or manual steps described so that the functionality of the software can be verified?
• Community guidelines: Are there clear guidelines for third parties wishing to 1) Contribute to the software 2) Report issues or problems with the software 3) Seek support

Software paper

• Summary: Has a clear description of the high-level functionality and purpose of the software for a diverse, non-specialist audience been provided?
• A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
• State of the field: Do the authors describe how this software compares to other commonly-used packages?
• Quality of writing: Is the paper well written (i.e., it does not require editing for structure, language, or writing quality)?
• References: Is the list of references complete, and is everything cited appropriately that should be cited (e.g., papers, datasets, software)? Do references in the text use the proper citation syntax?

Review checklist for @expectopatronum

Conflict of interest

• I confirm that I have read the JOSS conflict of interest (COI) policy and that: I have no COIs with reviewing this work or that any perceived COIs have been waived by JOSS for the purpose of this review.

Code of Conduct

• I confirm that I read and will adhere to the JOSS code of conduct.

General checks

• Repository: Is the source code for this software available at the repository url?
• License: Does the repository contain a plain-text LICENSE file with the contents of an OSI approved software license?
• Contribution and authorship: Has the submitting author (@martinju) made major contributions to the software? Does the full list of paper authors seem appropriate and complete?

Functionality

• Installation: Does installation proceed as outlined in the documentation?
• Functionality: Have the functional claims of the software been confirmed?
• Performance: If there are any performance claims of the software, have they been confirmed? (If there are no claims, please check off this item.)

Documentation

• A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
• Installation instructions: Is there a clearly-stated list of dependencies? Ideally these should be handled with an automated package management solution.
• Example usage: Do the authors include examples of how to use the software (ideally to solve real-world analysis problems).
• Functionality documentation: Is the core functionality of the software documented to a satisfactory level (e.g., API method documentation)?
• Automated tests: Are there automated tests or manual steps described so that the functionality of the software can be verified?
• Community guidelines: Are there clear guidelines for third parties wishing to 1) Contribute to the software 2) Report issues or problems with the software 3) Seek support

Software paper

• Summary: Has a clear description of the high-level functionality and purpose of the software for a diverse, non-specialist audience been provided?
• A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
• State of the field: Do the authors describe how this software compares to other commonly-used packages?
• Quality of writing: Is the paper well written (i.e., it does not require editing for structure, language, or writing quality)?
• References: Is the list of references complete, and is everything cited appropriately that should be cited (e.g., papers, datasets, software)? Do references in the text use the proper citation syntax?

[whedon]

Hello human, I'm @whedon, a robot that can help you with some common editorial tasks. @frycast, @expectopatronum it looks like you're currently assigned to review this paper 🎉.

⭐ Important ⭐

If you haven't already, you should seriously consider unsubscribing from GitHub notifications for this (https://github.com/openjournals/joss-reviews) repository. As a reviewer, you're probably currently watching this repository which means for GitHub's default behaviour you will receive notifications (emails) for all reviews 😿

To fix this do the following two things:

1. Set yourself as 'Not watching' https://github.com/openjournals/joss-reviews:

2. You may also like to change your default settings for this watching repositories in your GitHub profile here: https://github.com/settings/notifications

For a list of things I can do to help you, just type:

@whedon commands

For example, to regenerate the paper pdf after making changes in the paper's md or bib files, type:

@whedon generate pdf

[whedon]

Reference check summary:

OK DOIs

- 10.1007/s10115-013-0679-x is OK
- 10.1111/1467-9868.00125 is OK
- 10.1214/aoms/1177728190 is OK
- 10.18637/jss.v040.i08 is OK

MISSING DOIs

- None

INVALID DOIs

- None

[whedon]

👉 Check article proof 📄 👈

[terrytangyuan]

Note that @expectopatronum will not be able to review until Feb.

[frycast]

This package looks great. The vignette is especially detailed and useful.

I have two comments/questions:
NorskRegnesentral/shapr/pull/172 - A small typo here.
NorskRegnesentral/shapr/pull/171 - You may need to add the full MIT license here.

Thank you @martinju for making this package!

[nikolase90]

@frycast I've opened two issues based on your review.

• [review-frycast] Typo NorskRegnesentral/shapr#173
• [review-frycast] LICENSE NorskRegnesentral/shapr#175

[frycast]

@frycast I've opened two issues based on your review.

* [ ]  [NorskRegnesentral/shapr#173](https://github.com/NorskRegnesentral/shapr/issues/173)

* [ ]  [NorskRegnesentral/shapr#175](https://github.com/NorskRegnesentral/shapr/issues/175)

Thanks @nikolase90 for resolving both of those issues, and thanks for writing this great package along with @martinju .

I've opened one more small issue:

NorskRegnesentral/shapr#176

[nikolase90]

Hi @frycast; NorskRegnesentral/shapr#174 and NorskRegnesentral/shapr#177 are now merged into master. Let us know if you have any further comments, and thanks for the nice comment about the package!

[frycast]

Hi @frycast; NorskRegnesentral/shapr#174 and NorskRegnesentral/shapr#177 are now merged into master. Let us know if you have any further comments, and thanks for the nice comment about the package!

Thank you @nikolase90 . I've ticked off my review checklist, and now we wait to hear from @expectopatronum .

[martinju]

Hi @frycast; NorskRegnesentral/shapr#174 and NorskRegnesentral/shapr#177 are now merged into master. Let us know if you have any further comments, and thanks for the nice comment about the package!

Thank you @nikolase90 . I've ticked off my review checklist, and now we wait to hear from @expectopatronum .

Thanks for finishing the helpful review, @frycast .
I've now closed the license issue/PR.

[terrytangyuan]

Hi @expectopatronum, would you be able to start the review soon?

[expectopatronum]

Hi @expectopatronum, would you be able to start the review soon?

@terrytangyuan Yes, I am working on it!

[expectopatronum]

Hi @martinju,
there are some minor remarks regarding the references:

• Lundberg, S. (2019): I am not sure if this is correct/necessary. On their Github page they state which references should be used, this one doesn't seem to be there. (in case this actually should be there, I think SHAP should be capitalized)
• Maybe this also needs to be cited (very recent): https://raw.githubusercontent.com/slundberg/shap/master/docs/references/tree_explainer.bib
• Pedersen & Benesty: Title should be "lime: Local Interpretable Model-Agnostic Explanations"

Furthermore, capitalization of the titles and conference names is not consistent (some are capitalized, others are all lower case).

[martinju]

Hi @expectopatronum

Thank you for raising these issues.

I've opened an issue for these at NorskRegnesentral/shapr#178 (let's do any further discussion on there) and an associated PR at NorskRegnesentral/shapr#179

The pdf for the PR-version of the paper https://github.com/NorskRegnesentral/shapr/blob/references/inst/joss_paper/paper.pdf

Please let me know if you find this satisfactory.

[expectopatronum]

Regarding community guidelines: Could you add a hint where questions should be put? (most likely as a Github issue?)

[expectopatronum]

Regarding your vignette (part 1 - to be continued tomorrow):

• Just a minor remark (since it is not an issue with your package but with devtools): If you want user to be able to use vignette(shapr), they need to install the package using devtools::install_github("NorskRegnesentral/shapr", build_vignettes = TRUE), because apparently devtools does not build vignettes automatically.
• Building the vignette failed because of missing requirment caret, you might need to add it to your requirements? Same will happen for xgboost (I had to install it in order to run your example from the README), and gbm.
• For opening the vignette the user first has to figure out it's name, naming it 'shapr' would make that easier. Or mention in the description that one has to call vignette("understanding_shapr").

[nikolase90]

Hi @expectopatronum!

We'll add a remark that if you want to actually want to build the vignette and README you'll have to slightly modify the installment procedure.

By doing

devtools::install_github("NorskRegnesentral/shapr", dependencies = TRUE)

you'll also install packages listed under Suggets in DESCRIPTION (where xgboost and caret is listed). By doing it the current way we avoid that the users must download all the different packages. The default value is dependencies = NA, which only installs the packages listed under Depends, Imports and LinkingTo. If you set dependencies = TRUE you'll also install the packages listed under Suggets.

[terrytangyuan]

@nikolase90 I just realized that there are many optional dependencies in DESCRIPTION. Are there any dependent packages that are worth being cited?

[nikolase90]

@terrytangyuan Most of the packages listed under Suggests are not necessary to cite. Though we could add a citation for RcppArmadillo (since that is part of the core of our package). In addition to this we could also cite the packages of the models we support natively.

[terrytangyuan]

@nikolase90 Sounds good.

[expectopatronum]

Vignette (part 2):

• The citations look "weird", I tried to figure out whether I build the vignette wrong or it needs changing.
  
  If it looks ok for you, then ignore this comment :) (maybe you know how I can compile it correctly)

• Content:

and that we want to explain the predictions for new sets of features.

shouldn't this be:

and that we want to explain the predictions for new sets of examples. ?

• Formatting of the following formula (again, if I built the vignette incorrectly, never mind)
  

• Formatting again and it probably should say KernSHAP instead of KerSHAP

Besides those remarks the vignette is really great and helpful!

After NorskRegnesentral/shapr#178, NorskRegnesentral/shapr#180 and NorskRegnesentral/shapr#181 are resolved I am good to go!

[martinju]

Thanks for the vignette comments @expectopatronum
We will improve the vignette installation instructions and look into the potential typo.
Btw, we also have a html version of the vignette at our pckdown site: https://norskregnesentral.github.io/shapr/articles/understanding_shapr.html

To actually build the vignette with install_github you also need the argument build_vignettes = T

There have been some changes in devtools/remotes regarding defaults that are relevant if you want to build the vignette, but for versions newer than 2.1.0 and older than 2.0.0 (see r-lib/remotes#353), you should install the package with

devtools::install_github("NorskRegnesentral/shapr",dependencies = T,build_vignettes = T)

if you want to install the package (and build the vignette at installation)
To display the vignette, then simply run
vignette("understanding_shapr",package="shapr")
Does this solve your formatting problem?

[nikolase90]

@terrytangyuan NorskRegnesentral/shapr@5901393 and NorskRegnesentral/shapr@13f1d46 adds citations for additional packages.

[whedon]

Reference check summary:

OK DOIs

- 10.1007/s10115-013-0679-x is OK
- 10.1111/1467-9868.00125 is OK
- 10.1214/aoms/1177728190 is OK
- 10.18637/jss.v040.i08 is OK
- 10.1038/s42256-019-0138-9 is OK
- 10.1016/j.csda.2013.02.005 is OK
- 10.18637/jss.v077.i01 is OK

MISSING DOIs

- None

INVALID DOIs

- None

[terrytangyuan]

At this point could you make a new release of this software that includes the changes that have resulted from this review. Then, please make an archive of the software in Zenodo/figshare/other service and update this thread with the DOI of the archive? For the Zenodo/figshare archive, please make sure that:

• The title of the archive is the same as the JOSS paper title
• That the authors of the archive are the same as the JOSS paper authors
  I can then move forward with accepting the submission.

[nikolase90]

@terrytangyuan New release made with NorskRegnesentral/shapr#194. See also https://github.com/NorskRegnesentral/shapr/releases/tag/v0.1.1.

[terrytangyuan]

@whedon set v0.1.1 as version

[whedon]

OK. v0.1.1 is the version.

[terrytangyuan]

Thanks. Now we just need a DOI of the archive.

[nikolase90]

@terrytangyuan DOI: 10.5281/zenodo.3641831

[terrytangyuan]

@whedon set 10.5281/zenodo.3641831 as archive

[whedon]

OK. 10.5281/zenodo.3641831 is the archive.

[terrytangyuan]

@whedon accept

[whedon]

Attempting dry run of processing paper acceptance...

[whedon]

Reference check summary:

OK DOIs

- 10.1007/s10115-013-0679-x is OK
- 10.1111/1467-9868.00125 is OK
- 10.1214/aoms/1177728190 is OK
- 10.18637/jss.v040.i08 is OK
- 10.1038/s42256-019-0138-9 is OK
- 10.1016/j.csda.2013.02.005 is OK
- 10.18637/jss.v077.i01 is OK

MISSING DOIs

- None

INVALID DOIs

- None

[whedon]

Check final proof 👉 openjournals/joss-papers#1271

If the paper PDF and Crossref deposit XML look good in openjournals/joss-papers#1271, then you can now move forward with accepting the submission by compiling again with the flag deposit=true e.g.

@whedon accept deposit=true

[terrytangyuan]

@openjournals/joss-eics The paper looks good now. Could you take it from here?

[kthyng]

Ok things are looking good!

[kthyng]

@whedon accept deposit=true

[whedon]

Doing it live! Attempting automated processing of paper acceptance...

[whedon]

🐦🐦🐦 👉 Tweet for this paper 👈 🐦🐦🐦

[whedon]

🚨🚨🚨 THIS IS NOT A DRILL, YOU HAVE JUST ACCEPTED A PAPER INTO JOSS! 🚨🚨🚨

Here's what you must now do:

0. Check final PDF and Crossref metadata that was deposited 👉 Creating pull request for 10.21105.joss.02027 joss-papers#1272
1. Wait a couple of minutes to verify that the paper DOI resolves https://doi.org/10.21105/joss.02027
2. If everything looks good, then close this review issue.
3. Party like you just published a paper! 🎉🌈🦄💃👻🤘

Any issues? notify your editorial technical team...

[kthyng]

The DOI resolution is looking weird so let's wait before closing this up.

[kthyng]

@openjournals/dev This DOI link doesn't look like it usually does while it is resolving. Any ideas? Just wait longer?

[arfon]

This issue is being caused by formatted_doi: XX.XXXXX/joss.XXXXX being set in the paper.md. This shouldn't be in the template and a fix was applied in rticles in rstudio/rticles#272

I've reprocessed this submission manually and the updated PDF should be resolving soon on the JOSS website.

[whedon]

🎉🎉🎉 Congratulations on your paper acceptance! 🎉🎉🎉

If you would like to include a link to your paper from your README use the following code snippets:

Markdown:
[![DOI](https://joss.theoj.org/papers/10.21105/joss.02027/status.svg)](https://doi.org/10.21105/joss.02027)

HTML:
<a style="border-width:0" href="https://doi.org/10.21105/joss.02027">
  <img src="https://joss.theoj.org/papers/10.21105/joss.02027/status.svg" alt="DOI badge" >
</a>

reStructuredText:
.. image:: https://joss.theoj.org/papers/10.21105/joss.02027/status.svg
   :target: https://doi.org/10.21105/joss.02027

This is how it will look in your documentation:

We need your help!

Journal of Open Source Software is a community-run journal and relies upon volunteer effort. If you'd like to support us please consider doing either one (or both) of the the following:

• Volunteering to review for us sometime in the future. You can add your name to the reviewer list here: https://joss.theoj.org/reviewer-signup.html
• Making a small donation to support our running costs here: https://numfocus.org/donate-to-joss

[martinju]

I would like to express my gratitude to the editor @terrytangyuan and the two reviewers @frycast and @expectopatronum for a thorough, yet rapid, review process that led to an improved the paper, the package, and the descriptions in the README in our GitHub repo.

Thank you!

[kthyng]

Congrats to @martinju on your new publication! Thanks for your time and expertise to @terrytangyuan as editor and @frycast and @expectopatronum for reviewing.