Stef,

I think our banned troll is back :(  Beyond that, nothing as important as statistics is off-topic here.  I tend to use R because it does "everything," though its language is almost purely syntax over substance - hardly the OO bliss that its creators seem to believe it is.

Some things (e.g. linear regression, medians, quintiles in general) have filtered into my reach with Pharo.  Attached is some work so far on GSL, targeted at 1.3 and up.  You *might* find what you need in #gsl_stats_* methods, but no promises yet.  I regenerated the whole mess to use underscores, making it *far* easier to match code with the GSL documentation.  Ditto on the structures, which are a big mess right now.  I am trying to isolate "bad" things in packages marked "old" to ease ridding myself of Camel case where it hurts (interfacing).  Camel case is nice in wrappers, and I am using it there.

This will take time to finish, but I hope to release a set of wrappers that are important to me (YMMV) and the generated methods and structs to follow along in wrapping things important to others.  It is big library.

Challenges over time have included the long absence of callbacks, and various levels of underscore support that forced me to Camel case when I should not have done so, but had no choice "back when," and now problems passing void* and bumps in generating struct field accessors in recent FFIs.  Failing those things, this would have been done a while ago, but that's a lot of "ifs."

Bill