|
sxiao@SFWMD.GOV wrote:
>Sorry bothering you with one simple question.
That *is* why people answer posts here. Don't worry about it.
Besides, it might not be simple. I haven't seen anyone answer it yet.
> In proc univariate,
>probplot produces one pp-plot. But does anyone can hint me how to output
>the plot data?
>
>In other words, how to output the presumed theoretical reference data
>set?
I even looked in the SAS OnlineDoc (which is where you should be looking)
and I could not find an ODS table that holds this.
So you can just do it yourself. You have your X variable. Order them from
1 to n using PROC SORT. (Here 'n' is your sample size.) Then, for any
distribution F(x) for your values in order with i = 1 to n, you want
F-inverse
of
(i - 0.375) / (n + 0.25)
For the normal distribution, that's the PROBIT() function. Here's some
*untested* code showing you what I mean:
proc sort data=yourdata;
by x;
run;
data newdata;
set yourdata nobs=n;
quant = probit( (_n_-0.375) / (n + 0.25) );
run;
Now you plot X and QUANT. If you plot by percentiles of X, you
have a pp-plot.
>Following codes is what I used as an example (assuming normal dist.
>feature), and can anybody offer me why option "i=3Djoin" doesn't work in
>this case?
Oh, wait. You said you only had ONE question. :-)
>Statement "symbol" doesn't apply in proc univariate? But "v=3Ddot =
>cv=3Dred"
>seems working here. =20
>
>
>proc univariate data=3Dsashelp.class;
>
> var height;
>
> symbol1 i=3Djoin v=3Ddot ci=3Dblue cv=3Dred;
>
> probplot /normal(mu=3Dest sigma=3Dest color=3Dgreen w=3D2);
>
>run;
Okay, you cannot assume that SAS/GRAPH statements work here.
Look up the options that are available for the PROBPLOt statement
in PROC UNIVARIATE. You'll see that not all the SYMBOL statement
options are available. Or even apply. The SAS OnlineDoc has the
full details for the legal options.
HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330
_________________________________________________________________
Don’t just search. Find. Check out the new MSN Search!
http://search.msn.click-url.com/go/onm00200636ave/direct/01/
| 
