Check package with old publication dates

[tuxette]

cran_db <- tools::CRAN_package_db()
missingdata <- ctv::read.ctv("MissingData.md")
md_pkg <- cran_db[cran_db$Package %in% missingdata$packagelist$name, ]
# last publication is older and 2019 (2 years)
last_publication <- as.numeric(substr(md_pkg$Published, 1, 4))
md_pkg$Package[last_publication <= 2019]

• [X] accelmissing [OK, very specific application, complete, coming with a publication]
• [X] alleHap [OK, very specific because based on observed genotypes of parents]
• [X] BaBooN [OK, spécific and also referenced in Bayesian TV]
• [ ] brlrmr [covered, at least partially by mdmb) and not active]
• [ ] cassandRa [only one release, not active]
• [X] cat [old package, with reverse import]
• [ ] cdparcoord [no activity, partially covered in other packages for visualization or longitudinal data]
• [x] CircSpaceTime [OK because very specific but to re-check: only one release]
• [x] CoImp [OK because not covered elsewhere and cited in OfficialStatistics]
• [X] CRTgeeDR [OK because not covered]
• [x] cutoffR [OK because very specific but to re-check: only two releases]
• [ ] CVThresh [not maintained, broken URL, primary focus not on missing data]
• [ ] DiffusionRimp [package and author inactive since 2016]
• [ ] dils [only two releases; not active since 2013]
• [x] DrImpute [OK because has reverse dependencies]
• [x] DTWBI [OK: specific and associated to a recent publication but URL is dead]
• [x] DTWUMI [OK: specific and associated to a recent publication]
• [X] ECLRMC [OK: ensemble based, unique]
• [x] eigenmodel [OK, only one somewhere between regression and network]
• [x] experiment [OK, experimental design, unique]
• [ ] ForImp [not very specific, unmaintained]
• [x] FSMUMI [OK: specific and associated to a recent publication]
• [ ] GenForImp [removed: no activity since 2015, only one version]
• [x] GSE [OK: specific with several reverses]
• [ ] icdGLM [covered, at least partially by mdmb, and not active]
• [x] impimp [OK, only package for imprecise imputation with donors]
• [ ] imputePSF [one version only, not maintained]
• [x] imputeTestbench [OK benchmark dataset]
• [x] ipw [OK are package for IPW, many reverses]
• [x] IPWboxplot [OK are package for IPW]
• [X] ltm [more than 3000 downloads a month]
• [X] mi [more than 3000 downloads a month]
• [x] mimi [OK, low rank for mixed data]
• [X] missForest [more than 3000 downloads a month]
• [X] mitools [more than 3000 downloads a month]
• [x] mix [OK; one of the historic package]
• [X] norm [more than 3000 downloads a month]
• [x] pan [OK, same as mix]
• [x] phylin [OK, very specific with genetic distances]
• [x] prefmod [OK, only package for preference models]
• [x] pseval [OK, active and specific application fields]
• [ ] PSIMEX [no activity and the topic is covered by other packages, at least partially]
• [ ] PST [includes some functionnality on missing data but not the prior focus; inactive since 2017]
• [X] randomForest [more than 3000 downloads a month]
• [ ] reddPrec [no activity and partially covered elsewhere (binomial prediction)]
• [x] Rmagic [OK, has reverse dependencies in bioc]
• [x] rtop [OK, one of the main package for spatial interpolation]
• [ ] sbgcop [covered in mdgc, no activity]
• [x] scorecardModelUtils [OK, only package for scoring]
• [x] sievePH [OK, active, two articles]
• [x] stlplus [OK, historic package, probably still active]
• [x] StratifiedRF [unique for multi-table inputs]
• [x] swgee [OK, IPW for longitudinal data, not available elsewhere]
• [ ] TAR [one version, not active anymore since 2017]
• [x] tensorBF [only imputation method available for tensors as far as I can tell]
• [ ] ui [only two releases, both in 2019, not active]
• [ ] WaverR [only one version, not active since 2016]
• [ ] wNNSel [only one version, no updated since 2017]

[zeileis]

Thanks for checking this, @tuxette, this is very useful. I think you are doing this exactly the way that I would also do it, but I just wanted to comment to make sure: The old publication date alone is no reason for exclusion, I think. There are many packages that keep being useful where just no CRAN update was necessary. Also, if newer/better packages emerged this is no reason for excluding the established ones. But it may make a lot of sense to rephrase descriptions, e.g., by listing more powerful packages first and mentioning other packages only briefly.

[tuxette]

The old publication date alone is no reason for exclusion, I think

Sure! I'm checking them all, to distinguish between minor unmaintained packages and important well established packages. I think that the Missing Data task view currently has too many minor and outdated references. This is meant as a cleaning step.

But it may make a lot of sense to rephrase descriptions.

Do you mean the description in the task view itself or the short note that I'm writing (mostly for myself) alongside checked package names?

[zeileis]

I mean in the task view itself. For example, if you previously list only two minor packages method1 and method2 but now there is a comprehensive package manymethods. Then the old description might have been:

• Several methods are available for this task are available: method1 implements the standard method in this field while method2 provides a useful alternative for another specific setting.

But after discovering the manymethods package you would rephrase:

• The manymethods package provides a comprehensive set of methods for this tasking, include the classic standard method but also various alternatives. Some of these methods are also available in method1 and method2.

Thus, you wouldn't remove the old packages because they still fulfill the inclusion criteria for the task view. But by discussing the better and more comprehensive package in more detail you send a clear signal which package is more promising.

[tuxette]

finished in commit 0c777481ece85142616780b953547aa8b3cdc4e7

[zeileis]

Thanks! But I think that the replacement of "tables" with "les" was not intentional? See also the replacement of "established" with "es lished".

[tuxette]

Yes... :( that might have happened in other places. Wrong manipulation from my side. Thank you for pointed the errors: I'll correct them today.