Re: How to fit data with a straight line

[Steve Hankin <hankin@xxxxxxxxxxxxx>]

Hi Yang,

First I see a couple of more fundamental problems.

  1. The axis TAX is an year-by-year (non-modulo) axis. You have included a "modulo"
     qualifier, but I think it does not belong there.
  2. You have also asked for TEMP[gt=tax1@mod], modulo regridding. I believe what you
     want is simple TEMP[gt=tax1@AVE] -- to average the (monthly) source points into
     the (annual) destination cells.

Here is a complete session illustrating the type of calculation that you are
attempting. I use a SIN function for synthetic data. The resulting GIF file is
attached.

     yes? define axis/t=1-jan-1945:31-dec-1993:30/units=days tsrc
     yes? let my_var = SIN(T[gt=tsrc]/500)
     yes? plot my_var
     yes? def axis/t=1-jan-1945:31-dec-1993:365.25/units=days/edges tax1
     yes? let my_var_ann = my_var[gt=tax1@ave]
     yes? plot/over/symbols my_var_ann
     yes? go regresst

          ... Linear Regression Along the T Axis
          ... Instructions:
          Use the LET command to define new variables
          Define the variable P as your independent (X) variable
          Define the variable Q as your   dependent (Y) variable
          Results will be variables "SLOPE", "INTERCEP" and "RSQUARE"
          QHAT will be the regression estimate
          Note: If "T" is your independent variable then
          ...   "SET GRID Q" after defining Q.
          ...

     yes? let p = t[gt=tax1]
     yes? let q = my_var_ann
     yes? plot/over qhat
     yes? frame/file=qhat.gif

    Happy Ferreting - steve

================================================

Yang Haijun wrote:

> Hi,
>
>   Thank you very much. Bu now I have another problem,
>
>   Suppose I have data temperature with 49 year x 12 months = 588 time steps
>
>  currently SET data sets:
>     1> temp.nc  (default)
>  name     title                             I         J         K         L
>  TEMP     COADS SST                        1:97      1:57      ...       1:588
>  PERIOD   Averaging period                 ...       ...       ...       1:588
>
> 1)  First of all I get annual mean temp
>
>   yes? def axis/t=1-jan-1945:31-dec-1993:365.25/units=days/edges/modulo tax1
>   yes? let TMEAN = TEMP[gt=tax1@mod]
>
> 2)  Then I want to do regress for TMEAN
>
>   yes? let q = TMEAN
>   yes? let p = t
>   yes? set grid TMEAN
>   yes? go regresst
>
> Then I got error messages:
>
>  **ERROR: unknown defining grid: GT=TAX1@MOD --> g=user-or-pseudo-var not allowed
>  set grid TMEAN
>  Command file, command group, or REPEAT execution aborted
>
>  Obviously the problem is due to 't', how to give 't' to p here?
>  And let ferret know this 't' is corresponding to TMEAN?
>
>  Thanks a lot.
>
>
> Haijun Yang
>
>
>
>
> > From ferret_users-owner@ferret.wrc.noaa.gov Wed Mar 22 15:00:35 2000
> > From: "Mark Verschell" <verschell@neptune.gsfc.nasa.gov>
> > Date: Wed, 22 Mar 2000 15:45:57 -0500
> > In-Reply-To: Yang Haijun <navyang@tuna.meteor.wisc.edu>
>         "How to fit data with a straight line" (Mar 22,  1:48pm)
> > References: <200003221948.AA14779@tuna.meteor.wisc.edu>
> > X-Mailer: Z-Mail (5.0.0 30July97)
> > To: Yang Haijun <navyang@tuna.meteor.wisc.edu>,
>         ferret_users@ferret.wrc.noaa.gov
> > Subject: Re: How to fit data with a straight line
> > Mime-Version: 1.0
> > Content-Type: text/plain; charset=us-ascii
> > Sender: owner-ferret_users@ferret.wrc.noaa.gov
> > Precedence: bulk
> > Content-Length: 872
> >
> > Look at the .jnl files regress?.jnl that would be in $FER_DIR/go, since you are
> > trying to do this versus time, you would use regresst.jnl. A simple example for
> > your case:
> >
> > yes? let p=temperature
> > yes? let q=t
> > yes? set grid q
> > yes? go regresst
> >
> > Then:
> >
> > yes? list slope, intercep, rsquare ! slope is b and intercep is a
> >
> > You can also plot this with
> >
> > yes? plot qhat
> >
> > Mark
> >
> > On Mar 22,  1:48pm, Yang Haijun wrote:
> > > Subject: How to fit data with a straight line
> > > Dear ferret users,
> > >
> > > I have a time-series data, say Temperature with 40 years, now I want to fit
> > this data with a
> > > straight line, that is, I want to get this line:
> > >
> > >   Y = a + b*t
> > >
> > >  based on Temperature observations. Here Y is temperature, t is time step.
> > How do I get 'a' and
> > > 'b'? Is there a function to do this in ferret?
> > >
> > >  Thank you very much,
> > >
> > >  Haijun Yang
> > >-- End of excerpt from Yang Haijun
> >
> >
> >

--

                |  NOAA/PMEL               |  ph. (206) 526-6080
Steve Hankin    |  7600 Sand Point Way NE  |  FAX (206) 526-6744
                |  Seattle, WA 98115-0070  |  hankin@pmel.noaa.gov