Back in the 1980's, At&T produced S. Most statistical departments quickly worked with S; eg, Stanford, Purdue, Carnegie-Mellon. Numerous statistics departments contributed code to S. In the early 1990's, Statistical Sciences licensed S from AT&T, producing S-Plus. They put on a pretty front end, the sort of front end that gets you to buy Microsoft Word rather than use LaTeX. They charge about $1500 and produce it for most every platform but Linux.
Recently, the statistics departments, particularly one in Australia, began producing R, with a GNU like license. Some of the original developers for S now work on R. If you want the latest S-library for spatial statistics, you get it in R (December, 1997). And virtually all SPlus code works in R! No need for an SPlus GPL (analogous to Netscape's considerations); no need for SPlus. The Bureau of the Census uses R some because S has failed for some Census applications. SPlus told me a year ago and again two months ago that porting to Linux was not a priority for them. They might port Linux in 2 years. On the other hand, one developer of R stated that R has progressed so far and develops so quickly that SPlus may well become a minor player in the R & S arena. Today, we got a call that SPlus NOW HAS A VERSION FOR LINUX. They must have decided Linux is no minor player since developers of R often use Linux. As a Linux enthusiast, I'm glad to see the software, may even one day purchase it. However, I have been so satisfied with R that I'll probably follow Richard Stallman's wishes, using and contributing in minor ways to fine public software like R. I probably should say something about R. I use R on Debian Linux. After startup, you can immediately use it as a calculator (2+2=4). It solves most problems through vectors rather than pulling in a number at a time off disk. This makes it very quick, though it will balk on gargantuan data problems. It has cutting edge statistical procedures, including generalized-additive-models and loess-regression in addition to the standard anova, generalized-linear-models, ... . One of its fortes is graphics; eg, you can click a graphed point to get its name. For an R manual, just use an SPlus manual since the programming and most commands in SPlus work in R. To use either R or SPlus, you really do need a manual. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
