Rev 104 | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed
This is the README file for S-PLUS users. Some of it may be of interestto R users as well____________________________________________________________________This is a draft of version 5 of my survival package for S. Earlierreleases have appeared on Statlib (lib.stat.cmu.edu) as "survival","survival2", "survival3", and "survival4". The code found here shouldbe posted to statlib soon as well, after updating a few ".d" documentationfiles.There are 2 source trees, one for Splus version 5 and one for versions3 and 4. They come from a common code base, but the two differ in somedetails due to the new S4 computing engine. Use the appropriate one.Note that both contain important test directories "test", "testreg" and"testfrail".Changes that have occurred in the code since release 4 are documented inthe file "Changelog". The most important are-- many minor bugfixes. I say "minor" because most only happened withparticular sets of options that didn't occur very often.-- penalized models. See the frailty.ps file for more information-- [] methods. In particular, this gives more flexibility ingraphing curves.-- updated survreg code-- further plotting options for survival curves.Note that most of the changes are already a part of your distribution ofSplus. Bug fixes, in particular, are sent to Statsci as they are found.I expect all of the changes herein to be part of the shipped Splus distributionwithin a year. (Validation, documentation, and bundling into a release alltake time). The largest reason for using this code in the meantime is thepenalized models, which include random effects as an important special case.Before loading this try> is.loaded("coxfit4_a") # penalized Cox models included?> is.loaded("survreg4") # newest survreg functions ?if both are true then the "standard" release is more up to date than thisarchive.One area of lack in the enclosed are the help (.d) files. Statsci hasdone a fair bit of editing and improvement to the ones I first wrote, andmany (most?) of these changes have not been re-incorporated into my sourcecode. I don't expect that they ever will; this is an editorial taskthat I just can't get excited about.Please see the notes on expected survival, copyright, and the installdirections in each subdirectory.---------------------I don't use Splus on Windows, don't even have a copy of Splus for Windowsin house, and am not intending to buy one soon. I know that the procedurefor compiling/linking routines under Windows is different, but having neverdone it this page can offer no useful information on how to go about theprocess.