Re: Need multiple regression source/book

Richard Ulrich seemed to utter in news:u0t991pv7qqufa69rqr024ojh72k7hj88g@
4ax.com:

> On Wed, 25 May 2005 21:03:03 GMT, trw7at@xxxxxxxxxxxxxxxxxx (Tim
> Witort) wrote:
>
>> zhen seemed to utter in news:1117047530.325177.122480
>> @g43g2000cwa.googlegroups.com:
>>
>> > Tim,
>> >
>> > You might want to try Lapack/Blas port from Fortran to Java.
>> > COM/ActiveX are MS technologies, however. But if you're willing to go
>> > Java, then you can Lapack/Blas http://www.netlib.org/.
>> >
>> > You can do all the regression you want with tried/tested algorithms.
>>
>> Thanks for the link. I'll take a look. For whatever reason,
>> I seem to be having a hard time finding the tried/tested algorithms
>> on the net. I'm sure they are out there since statistics was one of
>> the first uses of computers. I must not be looking in the right
>> places or using the right search paramaters!
>
> I just googled on <computational-algorithms regression fortran >
> and got 426 hits. That's too many to be immediately useful.
> I put in "fortran" to reduce the number from many thousands.
> Still, that's something to scan.
>
> It occurs to me that a big part of the concern for robust
> algorithms has to do with problems like those presented
> by the Longley data - numbers like 10000001, 10000002, ...
>
> I get 86 hits for < regression algorithms longley-data > .
>
> That problem for the computer program is mainly prevented
> by *not* using the simple [summation X, summation
> X-squared] formula which is in all basic textbooks.
>
> Hope this helps.

That is helpful, thanks. I'll have to brush up on my Fortran!

-- TRW
_______________________________________
t r w 7
at
i x dot n e t c o m dot c o m
_______________________________________
.

Relevant Pages

  • Re: custom symbol for derived type component selector?
    ... For today, the answer is no, not in standard Fortran. ... accept a dot, but that is not standard and I strongly recommend against ... Using that nonstandard feature will cause you a lot more ... potential conflicts with other uses of the dot. ...
    (comp.lang.fortran)
  • Re: ReNewAmerica: Another loony theory from the wonderland of
    ... A computer program requires a programmer, ... requires a designer, is perfectly logical. ... My rejection of evolution has everything to do with rational thinking ... email to oshea dot j dot j at gmail dot com. ...
    (talk.origins)
  • Re: should i write keywords uppercase?
    ... I still am comfortable have all Fortran statements in upppercase ... I don't speak for Intel, ... Ken Fairfield ... kenneth dot h dot fairfield ...
    (comp.lang.fortran)
  • Re: gfortran bug?
    ... Try this with Compaq Fortran, then buy Steve Lionel a crate of beer. ... qomputing dot demon dot co dot uk ... > from the way that I declared it in my interface block. ... > and compile it with my GNU Fortran 95 compiler: ...
    (comp.lang.fortran)
  • Re: please let me know what this program prints
    ... Bob Corbett ... HP Fortran 77 V7.6-198-48D52 on OpenVMS Alpha V7.3-1 ... (This is an F95 compliant compiler so according to my ... kenneth dot h dot fairfield ...
    (comp.lang.fortran)