Hello:
I am wondering if anyone out there has implemented routines
in Lisp that perform linear / mulitple regressions. If possible, I
would really appreciate if someone could E-mail or post such routines.
My E-mail address is ········@mcs.drexel.edu.
Thank you
David J. Krassen
In article <······················@mcs.drexel.edu> ········@mcs.drexel.edu ( David Krassen) writes:
>
> I am wondering if anyone out there has implemented routines
>in Lisp that perform linear / mulitple regressions. If possible, I
>would really appreciate if someone could E-mail or post such routines.
>My E-mail address is ········@mcs.drexel.edu.
>
>Thank you
>David J. Krassen
>
This is not quite what you are looking for, but it may help. I have
put together a statistical analysis system based on lisp, xlisp in
particular. The basics for linear regression are written in C and then
used from the Lisp level. Here is a copy of a blurb on the system:
-------------------------------------------------------------------------------
This note is to announce the availability of Version 2.1 of
XLISP-STAT, an extensible system for statistical computing and dynamic
graphics.
XLISP-STAT is based on a dialect of Lisp called XLISP. It is available
for the Apple Macintosh, for workstations running the X11 window
system under BSD UNIX (and perhaps System V with BSD enhancements),
and for Sun workstations running SunView. A version for the Commodore
Amiga is available as well.
The system is based on David Betz' XLISP 2.1, which is written in C.
To facilitate statistical computations I have modified standard Lisp
functions for addition, logarithms, etc., to operate on lists and
arrays of numbers, and have also added a number of basic statistical
functions. About two thirds of the additions are written in C; the
rest are written in Lisp. Several basic forms of plots, including
histograms, scatterplots, 3-dimensional rotatable plots and
scatterplot matrices are provided. These plots support various forms
of interactive highlighting operations and can be linked so points
highlighted in one plot will be highlighted in all linked plots.
Interactions with the plots are controlled by the mouse, menus and
dialog boxes. An object-oriented programming system is used to make it
possible to customize menus, dialogs, and the way plots respond to
mouse actions.
Source code for the UNIX version, which can be compiled without
graphics, for X11 graphics, or for SunView graphics, is available by
anonymous ftp from
umnstat.stat.umn.edu (128.101.51.1)
It is in the file xlispstat2.1R1.tar.Z in the directory pub/xlispstat.
(The version and release numbers will be updated as changes are
made/bugs are fixed). This system has been compiled on a VAX, DEC
3100 (pmax), Sun 3, Encore Multimax and a Cray XMP.
The Macintosh version is also available by anonymous ftp from the
address given above. The Macintosh distribution consists of two parts,
contained in the files
XLISP-STAT2.1R1.Disk1.sit.Hqx
XLISP-STAT2.1R1.Disk2.sit.Hqx
in the directory pub/xlispstat. These files contain archives created
with StuffIt 1.5.1 that have been binhexed with BinHex 4.0. The
Macintosh distribution includes two versions of the system, one for
macs with a math coprocessor and one for generic macs. Source code for
the Macintosh version is included in the UNIX sources.
For reasonable operation the Mac version requires 2MB of memory. It
may be possible to use it on very small problems with only 1MB. The
program is fairly slow on a Mac Plus or SE.
A tutorial introduction to the system is available. It is written
primarily for the Macintosh version, but the differences to the UNIX
version are minor. The tutorial is available as a set of LaTeX files
in xlispstat.doc.tar.Z in the directory pub/xlispstat at the ftp
address above.
If you do not have access to anonymous ftp, you can obtain the source
code and documentation from the statlib archive. To find out how to do
this, send a mail message to
·······@lib.stat.cmu.edu
containing the line
send index from xlispstat
More complete documentation is available as a book published by Wiley.
The title of the book is "Lisp-Stat: An object-oriented environment
for statistical computing and dynamic graphics."
Information on Current LISP-STAT Implementations
================================================
The most recent version of xlispstat is 2.1.R2.
Implementations are available for the Macintosh, UNIX systems running
X11, Sun workstations running SunView, and for the Commodore Amiga.
At this time there is no version available for DOS or OS/2. New
versions or releases will be announced to the sci.math.stat network
news group.
The source code for the UNIX and Macintosh versions is available for
anonymous ftp from umnstat.stat.umn.edu (128.101.51.1) in directory
pub/xlispstat and from the statlib mail archive. To obtain the source
from statlib, start by sending an email message to
·······@lib.stat.cmu.edu with the single line
send index from xlispstat
and follow the instructions you receive in reply.
Executables for the Macintosh are available for anonymous ftp from
umnstat.stat.umn.edu as BinHexed StuffIt files. If you do not have
access to anonymous ftp over the internet, you can obtain the
Macintosh version by sending a) two formatted 800K Macintosh disks and
b) a stamped, addressed diskette mailer to
Lisp-Stat Information
School of Statistics
270 Vincent Hall
University of Minnesota
Minneapolis, MN 55455
If you do not have access to US postage stamps, you can include
international reply coupons.
Executables for the Amiga are available for anonymous ftp from
umnstat.stat.umn.edu in directory pub/xlispstat/amiga. Executables and
source code are also available in the Fish collection on Fish disks
385 and 386. The contents of these disks is also available by
anonymous ftp from ux1.cso.uiuc.edu (128.174.5.59).
NOTES
=====
A simple system for fitting generalized linear models in xlispstat is
available for anonymous ftp from umnstat.stat.umn.edu or by email from
statlib. This system is included on the second disk of the Macintosh
distribution.
From: Richard A. O'Keefe
Subject: Re: Lisp/Regression Routines
Date:
Message-ID: <5851@goanna.cs.rmit.oz.au>
In article <······················@cs.umn.edu>, ····@umnstat.uucp (Luke Tierney) writes:
> I have put together a statistical analysis system based on lisp,
> xlisp in particular.
I suppose this is a lot to ask for, but does it resemble Xerox's
"Interactive Data-analysis Language" (IDL) at all? (That was built
in Interlisp-D; no C required or possible.)
--
There is no such thing as a balanced ecology; ecosystems are chaotic.