Steve and Jami,

For some reason I don't have the whole thread of this message...

If you can deal with the fact that SQL Server is a little less precise
about numerical methods when dealing with huge volumes of data,
most of the standard statistical functions are not that complicated to
program, and those that are less simple can sometimes be implemented
with an auxiliary table - such as for normal distribution or t-tests. For
example, see my examples of least squares fit, skew, kurtosis, and
covariance at [url]http://www.users.drew.edu/skass/sql[/url]. I don't have any
examples posted that use a table of z- or t- scores, but a good statistics
text should help.

Some things are probably not so easy, like logistic regression, but
if you look into the mathematics behind many of the standard statistics,
you'll find that sql's aggregate functions take you a long way.

Steve Kass
Drew University

Steve Brokaw wrote:
>I'm also trying to do the same thing, actually only a few
>key statistical functions. I checked into the IMSL option
>described below but I'm not sure that will work. We have
>several SAS users here, but trying to integrate SAS
>calculations into my processes would be messy. And SAS is
>EXTREMELY expensive, and cost continues year after year.
>
>Wondering if you ever found any good solutions. I will
>post or email if I find anything.
>
>
>
>>-----Original Message-----
>>Yes, I'm aware of SAS and SPSS. But I am looking for
>>
>>
>something that
>
>
>>integrates natively with MS SQL Server. I would like to
>>
>>
>manage all of my
>
>
>>data scrubbing, production, and wharehousing within the
>>
>>
>same environment. -
>
>
>>Jami
>>
>>
>>"Pele" <rog11228aol.com> wrote in message
>>news:Oq7f$ut$CHA.2184TK2MSFTNGP11.phx.gbl...
>>
>>
>>>"Jami Albro-Fisher" <jamifazzi.com> wrote in message
>>>news:3e94688d$1_6corp-goliath.newsgroups.com...
>>>
>>>
>>>>I need to perform various statistical calculations on
>>>>
>>>>
>data in SQL
>
>
>>Server.
>>
>>
>>>>These include frequency distributions, variance,
>>>>
>>>>
>correlation,
>
>
>>chi-square,
>>
>>
>>>>t-test, and others. But I'm suprised at the dearth
>>>>
>>>>
>of resources
>
>
>>available
>>
>>
>>>>for this purpose. Is Total SQL Statistics the only
>>>>
>>>>
>option? I am more
>
>
>>>>interested in UDF's and stored procs because
>>>>
>>>>
>performance is an important
>
>
>>>>requirement. Are their open source or commercial
>>>>
>>>>
>options available?
>
>
>>Any
>>
>>
>>>>information or leads would be much appreciated! TIA.
>>>>
>>>>
>>>Commerical systems to consider are SAS (Statistical
>>>
>>>
>ysis System)
>
>
>>>[url]www.sas.com[/url]
>>>Superior software for statistical ysis,data
>>>
>>>
>manipulation,sql and much,
>
>
>>>much more.But expensive.
>>>SPSS ([url]www.spsss.com[/url]) should be able to do everything
>>>
>>>
>you want
>
>
>>>and more.
>>>You can also check out the IMSL library.A library of
>>>
>>>
>stat/math routines
>
>
>>>in fortran and C.Search google for url.
>>>
>>>
>>>
>>>
>>
>>
>>-----------== Posted via Newsfeed.Com - Uncensored Usenet
>>
>>
>News ==----------
>
>
>> [url]http://www.newsfeed.com[/url] The #1 Newsgroup Service
>>
>>
>in the World!
>
>
>>-----= Over 100,000 Newsgroups - Unlimited Fast
>>
>>
>Downloads - 19 Servers =-----
>
>
>>.
>>
>>
>>