🚀 go-pugleaf

Gary Perlman's UNIXSTAT is available on many BBSs all around the country.
I has both source (in C) and documentation.

In article <874@bingvaxu.cc.binghamton.edu> vu0112@bingvaxu.cc.binghamton.edu (Cliff Joslyn) writes:
>
>I'm looking for a good, basic statistics package for the PC, ideally a C
>library/source. I have enough experience w/SAS on VAX to know I hate it.
> I need basic regression, ANOVA, many forms of hypothesis testing.

A very good package for the PC which is not C but APL is Statgraphics.
By a company called STSC (?).  It is not the easiest program in the
world to use but it is very good.  The application does ANOVA, curve 
fitting(polynomial, spline, and others), smoothing, quadrature,
and many other types of numerical analysis.

If you would like more info please EMail me at wasp.pawl.rpi.edu.
				Alexander J. Denner
				320 Crockett Hal
				R.P.I.
				Troy, N.Y. 12180

I used Statgraphics for a while and did not like it. I don't
have my notes but I will summerize what I can remember. Also,
some of the bugs and design errors may now be fixwd.

It is a pure menu driven package with no alternate command
language or macros. Very tediuos to use for large problems.

It has the worst transformation system that I have ever used.
It might be OK if you junked it and used the APL link. No IF
statement, many bugs, flaws and design glitches.

The "spreadsheet" style data editor is too slow for practical
use.

The automatic graph scaling routines are poor. No reason
for this in a graphics package. Limited control of graphs.

Bottom interval of histogram is ommited.

etc.

I finnally gave up on using it and switched to Stata, which
leaves a lot to be desired but light years ahead of
statgraphics for real problems. I have to supplement stata
with Systat to get a reasonable range of statistical methods.

Don Martin, Dept. of Biostatistics, Univ. of WA.

Thanks for all the advice on stats packages, folks.  I eventually found
a copy of UNIX-STAT on my friendly local *LTRIX machine, and it has made
me happy.  No chi-square or autocorrelation though.  Sob.

O---------------------------------------------------------------------->
| Cliff Joslyn, Professional Cybernetician 
| Systems Science Department, SUNY Binghamton, New York, but my opinions
| vu0112@bingvaxu.cc.binghamton.edu
V All the world is biscuit shaped. . .

I am not sure what the original poster wanted in a stat package, but
there are several professional packages available. If you do time
series (as I do) then I strongly recommend RATS (version 2.12)
available from VAR Econometrics, PO Box  1818, Evanston, IL
60204-1818 (312 864 8772). For cross-sectional work, try SST
(Dubin-Rivers, Pasadena, CA). Both are quite profession, and while
neither is remarkably user-friendly, I can't see why anyone who
would
want to use these packages would have any problems.

If you want to do non-standard things, the only way to go is GAUSS
(Aptech
in Seattle - Lake Union Place Suite 301, Seattle, 98103 - 206 547
1733

Gauss is fabulous - you can do anything, and while it is a bit like
programming, they have designed modules to make it very easy to do
the
most complicated maximum likelihood routines. For example, I got my
Kalman filtering programs running (that is providing estimates)in 
under half a day. 

Compared to these programs, things like STATA and Statgraphics are
quite amateurish (though they provide much of what many people
need).

I have no connection with VAR, Dubin-Rivers or APtech, except as a
very (well pretty) satisfied user.

Neal Beck
Dept. of Pol Sci
UCSD

In article <3640@sdcc6.ucsd.EDU> ir1@sdcc6.ucsd.edu.UUCP (Nathaniel L. Beck) writes:
>
>I am not sure what the original poster wanted in a stat package, but
>there are several professional packages available. If you do time
>[recommendations of rats, sst and gauss deleted]
>
>Neal Beck
>Dept. of Pol Sci
>UCSD

i also do not know what the original poster asked, but s/he may also
look at micro-tsp 5.1 (quantitative micro software, 4521 campus drive, suite
336, irvine, ca 92715 [714]856-3368).  the software is very good for
time series, but would also be satisfactory for cross sectional designs
(it includes weighted procedures, for example).  it does not presently
have the ability to handle pooled data sets easily.  utsp will do ols,
2sls and systems of equations, and does n-th order autocorrelation and
moving averages, as well as full arima.  it has good transformations and
import/export, and satisfactory plotting and graphing.  the user
interface is much friendlier than rats, i'm told by a user of both.
utsp also includes rudimentary data entry and editing abilities.  this
is just a sketch, not mentioning :-) logit and nonlinear estimating
ability...

the package is $595, or $295 education discount.  it is, alas, copy
protected.
-- 
 Frank C. Burleigh
 Institute for Social Research - Indiana University - Bloomington, IN 47405
 ihnp4!inuxc!isrnix!frank
 {pur-ee,kangaro,iuvax}!isrnix!frank