Code graveyard

[cwolock]

This is where we'll discuss code that can be dropped from the package.

[cwolock]

clusterStatsOld

[cwolock]

Pretty self-explanatory. There's a well-documented version called clusterStats. Whether or not we decide to keep that one, clusterStatsOld seems superfluous.

[cwolock]

I deleted this function. I compared it with clusterStats and it has strictly less functionality.

[cwolock]

correlate

[cwolock]

This function computes a correlation matrix for an arbitrary number of variables, optionally within strata. It also provides inference, including t-tests and confidence intervals.

[cwolock]

tabulate

[cwolock]

This function creates a contingency table for categorical variables and computes ORs, RRs, chi-squared statistics, etc. I believe it's similar to the epi.2by2 function in epiR that @jphughes9 used last year in 511

[jphughes9]

I'm pretty happy with the epiR and epi.2by2 except it doesn't have the McNemar test ... but then neither does tabulate.

[cwolock]

I'll add that this function is mostly a wrapper of tableStat, which is itself mostly a wrapper of descrip.

[cwolock]

tableStat

[cwolock]

Like tabulate, relies heavily on descrip. I think (having not really found specific use cases for this function) that it's mostly a wrapper for descrip, allowing users to choose subsets of descriptive statistics and providing some sort of nicer formatting.

[cwolock]

scatter

[cwolock]

Basically plot with certain defaults (like jitter) and superimposed loess/least squares lines.

[cwolock]

oneSample

[cwolock]

This is a big one. It's similar in spirit to regress, but for a single variable. Can perform inference for means, proportions, odds, survival probabilities, etc. As far as I can tell, for certain functionality (like a basic t-test) it's redundant with ttest, but the output is a bit less detailed. This function certainly seems useful in theory. I'd be curious to hear if/how people use it in practice.

[cwolock]

clusterStats

[cwolock]

Another wrapper for tableStats that provides within-cluster descriptive statistics. Also does within-cluster least squares slopes.

[jphughes9]

Insofar as 511 is concerned we only do one sample t-tests and proportion tests. ttest() already covers the “mean” option and we talked about developing functions for proportion tests (equivalent of prop.test and binom.test with better output) so I don’t see a need for oneSample from the 511 point of view. Jim -------------- Jim Hughes Professor of Biostatistics University of Washington Email: ***@***.******@***.***> Mailstop 351617 Phone: 206-616-2721 Seattle, WA 98195 Fax: 206-616-2724 Cell: 206-498-2456 From: Charles Wolock ***@***.***> Sent: Saturday, August 21, 2021 10:22 PM To: statdivlab/rigr ***@***.***> Cc: Jim Hughes ***@***.***>; Mention ***@***.***> Subject: Re: [statdivlab/rigr] Code graveyard (#49) This is a big one. It's similar in spirit to regress, but for a single variable. Can perform inference for means, proportions, odds, survival probabilities, etc. As far as I can tell, for certain functionality (like a basic t-test) it's redundant with ttest, but the output is a bit less detailed. This function certainly seems useful in theory. I'd be curious to hear if/how people use it in practice. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#49 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AFPGUCWHHKQZITAKZDEXI4TT6CCQHANCNFSM5BU2NGTQ>. Triage notifications on the go with GitHub Mobile for iOS<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android<https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email>.

[taylorokonek]

lspline and lsplineD

[taylorokonek]

We've decided to no longer support spline functionality in regress, and all functionality for splines has been removed from regress. The functions lspline and lsplineD still exist however in their own scripts, and we can likely delete them.