I have just uploaded to CRAN the new version of the randomForest package. Those who had used the previous version will absolutely want to upgrade, as the new version fixes a few serious bugs (some giving wrong results w/o warning or error). This version matches version 3.3 of the classification code and version 1.0 of the regression code on Prof. Breiman's web site. The package provides an R interface to Prof. Breiman's code for regression and classification by the random forest algorithm. Details about the algorithm (and the original Fortran code) can be found on http://oz.berkeley.edu/users/breiman New features and bug fixes include: o Random forest for regression problems now incorporated into the package. o Factor (categorical) predictors are now handled correctly, especially through the formula interface. (The previous version wrongly expands factors into contrasts.) o Ordered factor predictors are now correctly handled. o For classification problem, there is an new option for the predict method to calculate proximity measure for the test data. o Calls to R's random number generator from Fortran have been made more portable. I believe this was causing segmentation faults on some platforms in the previous version. o The Example section of the help page for randomForest now show some of the extra things that can be done with the randomForest object. o The call to the underlying C/Fortran code no longer copy data (i.e., .C(..., DUP=FALSE) is used). This should reduce memory requirement quite a bit, especially for large data sets. This should be fairly safe since data pass to C/Fortran are copies made at the R level anyway. o The version number of the package now tracks the version number of Prof. Breiman's classification code (i.e., currently 3.3). o The spelling of my co-author's name has been corrected! Because of the limited amount of time I was able to spend on this package, the code are not the most elegant. I would appreciate any comments/suggestions/bugs/patches. Regards, Andy Andy I. Liaw, PhD Biometrics Research Phone: (732) 594-0820 Merck & Co., Inc. Fax: (732) 594-1565 P.O. Box 2000, RY84-16 Rahway, NJ 07065 mailto:andy_liaw at merck.com ------------------------------------------------------------------------------ Notice: This e-mail message, together with any attachments, contains information of Merck & Co., Inc. (Whitehouse Station, New Jersey, USA) that may be confidential, proprietary copyrighted and/or legally privileged, and is intended solely for the use of the individual or entity named on this message. If you are not the intended recipient, and have received this message in error, please immediately return this by e-mail and then delete it. ============================================================================= -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-announce mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-announce-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._