Re: multiple regression

From: Mateus Teixeira <mateus.teixeira_at_nyahnyahspammersnyahnyah>
Date: Wed Oct 03 2012 - 13:37:24 MDT

Hi,

Well, I think so. In that way, you will have 3 partial coefficients for
each grid point.

However, *rc* variable should have its rightmost dimension equal to the
number of partial coefficients of this multiple regression, e.g, 3. Also, I
suppose that 1680 is the number of time instants of your data (rightmost
dimension of *gsst*).

Best regards

Mateus

2012/10/3 95108084 <g9510808@ms2.pccu.edu.tw>

> Hi,
> Your mean like this?
> x = new((/3,1680/),float)
> x(0,:) = 1.0
> x(1,:) = nino12
> x(2,:) = nino4
> rc = gsst(lat|:,lon|:,time|:)
> printVarSummary(x)
> printVarSummary(gsstano)
> do i = 0,88
> do j = 0,179
> rc(i,j,:) = reg_multlin(gsstano(i,j,:),x,False)
> end do
> end do
>
> Many Thanks
>
>
> *On Wed, 3 Oct 2012 16:10:43 -0300, Mateus Teixeira wrote*
> > Hi,
> >
> > I think that you just need to organize your data to fit this function.
> As you can see in the description of this function, the lines of X array
> identifies the multiple variables that influence Y, a multiple linear
> regression.
> >
> > Are you interested in a multivariate regression? If so, this function
> isn't what you want, really.
> >
> > Best regards,
> >
> > Mateus
> >
> > 2012/10/3 95108084 <g9510808@ms2.pccu.edu.tw>
> >
>>
>> Hello,
>> > Does anyone know a function for performing a multiple regression with
>> NCL?
>> > I only find this
>> > http://www.ncl.ucar.edu/Document/Functions/Built-in/reg_multlin.shtml
>> > But this function just only for one-dimension matrix.
>> > I want to do on the 3-dimension matrix (global grib data).
>> >
>> > Many Thanks
>> >
>> > _______________________________________________
>> > ncl-talk mailing list
>> > List instructions, subscriber options, unsubscribe:
>> > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>> >
>
>
>
>
> --
> �����Ƥj��
> WebMail system
>
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Oct 3 13:37:35 2012

This archive was generated by hypermail 2.1.8 : Mon Oct 08 2012 - 15:54:16 MDT