Subject: Repost: SAS JMP - any opinions?
From: "Jack Duffy"
Date: 1 Dec 1996 12:51:06 GMT
My apologies for wasting bandwidth -- I omitted my e-mail address on 1st
post.
I'm a quality engineer evaluating Windows general purpose stat. software.
Is there anybody out there who has looked at SAS JMP? I downloaded their
demo and it looks quite nice. In the past, I've used Statgraphics for DOS
and BBN's RS1 and Cornerstone. I'm looking for decent EDA capability, the
usual litany of descriptive stats / graphs, and multiple regression and
anova. An array of time series tools would be a plus. Also, all flavors
of SPC and DOE that won't lead me astray. Please answer by e-mail and I'll
summarize responses & post to the group within a week or so, depending on
response rate.
Thanks........ Jack Duffy, Peoria IL......... DUFFYJE@CEDAR.CIC.NET
Subject: Re: Need cosine-distributed random help.
From: mcohen@cpcug.org (Michael Cohen)
Date: 1 Dec 1996 22:51:31 GMT
DR TE$TH & THE ELECTRIC MAYHEM (ronb@cc.usu.edu) wrote:
: I need to generate a series of random numbers whose distribution follows the
: cosine curve. I.e.
:
: |^^---..
: | ^^--.
: | ^^-.
: | ^-
: | ^-
: | ^.
: | -
: | -
: | -
: |-------------------------|
: 0 PI/2
:
: Each value from 0 to PI/2 must occur cosine (value) times as often as the
: frequency of the control value (zero).
:
: I have tried doing it the following way (C code):
:
: do {
: r=rand()/RAND_MAX; /*generates a random value from 0 to 1...*/
: } while (r==1.0); /*PI/2 has a probability of zero, so throw it out...*/
: angle=PI_OVER_2-acos(r); /*this _should_ give me what I'm after, as far as I
: can tell.*/
:
: If I run a histogram on the distribution of this function, I get something that
: looks very much like cosine. However, when I apply it in a program that I've
: written, I do not get correct results in my calibration routines.
:
:
We need the inverse (say , G) of the (cumulative) distribution function F.
From the diagram (or description), it seems that F(x)=sin(x), so G(y)=arc
sin(y). I suppose in C this would be asin(y). Therefore, if u is
uniformly distributed on [0,1], then asin(u) has the distribution sought.
--
Michael P. Cohen home phone 202-232-4651
1615 Q Street NW #T-1 office phone 202-219-1917
Washington, DC 20009-6331 office fax 202-219-2061
mcohen@cpcug.org
Subject: Re: Multiple comparisons
From: mcohen@cpcug.org (Michael Cohen)
Date: 1 Dec 1996 23:24:20 GMT
STEVEN N GOODMAN (sgoodman@WELCHLINK.WELCH.JHU.EDU) wrote:
: Stat-L'ers
: I am writing a commentary on the issue of multiple comparisons. I am
: well acquainted with the frequentist v. Bayesian perspectives, but I would
: like to read something that addressed this issue on a somewhat deeper
: epistemological level, i.e. what is the meaning/justification/significance
: of grouping classes of objects into "families". This also has some
: relevance to the combining of studies in meta-analyses, and to
: the identification of "clusters" in spatial statistics. I am sure there
: must be some good stuff on that, but I am rooting around in a
: self-referential literature which never directly addresses things on that
: level. Any suggestions? Thanks.
:
I suggest looking at Susan W. Ahmed's paper, "Issues arising in the
application of Bonferroni procedures in federal surveys", 1991
American Statistical Association Proceedings of the Survey Research
Methods Section, 344-349.
--
Michael P. Cohen home phone 202-232-4651
1615 Q Street NW #T-1 office phone 202-219-1917
Washington, DC 20009-6331 office fax 202-219-2061
mcohen@cpcug.org
