Date:         Thu, 2 Dec 2004 14:55:59 -0800
Reply-To:     cassell.david@EPAMAIL.EPA.GOV
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "David L. Cassell" <cassell.david@EPAMAIL.EPA.GOV>
Subject:      Re: gplot question
In-Reply-To:  <200412020118.iB21I4GF001122@listserv.cc.uga.edu>
Content-type: text/plain; charset=US-ASCII

Ben <benpub7@YAHOO.COM> replied:
> thanks, ya huang. it is one solution.
>
> but what I intended to get is kind of like regression lines,  for
every x
> values, the y has percentiles, they are functions of x,
>   p10=f1(x), p20=f2(x), p50=f3(x) .......

Okay, I'm still not clear on what you really want.  It seems that
you want percentiles for all values of X, turned into additional
contours on the plot.  But how do you plan to interpret P10 for the
lowest value of X, given that you are thereby (implicitly) extrapolating
out past the range of X in order to do so?  Or do you have a large
number of Y values for every X, in which case you could actually
generate boxplots for each X using PROC BOXPLOT ?

One simple alternative you might want to look into is in the SYMBOL
statement.  You can specify regression curves using INTERPOL=RxCLMyy
where x is L (for a linear fit) or Q (for quadratic) or C (for cubic),
CLM gives confidencel imits for the mean regression line, and
yy is a number between 50 and 99 that represents the confidence level.
yy=95 is typical.

HTH,
David
--
David Cassell, CSC
Cassell.David@epa.gov
Senior computing specialist
mathematical statistician