Hi, all, I want to use lowess(x, y) where x and y are vectors of length of 4000+. In fact, x and y are log of some vectors. So, some of the elements are NaN. lowess() can not take away those elements then do the fitting. It will give the error message and do nothing. 1. Can anybody tell me how to get rid of those NaN's and use lowess()? 2. How to get the LOWESS fitting values for any elements in x? Thank you in advance. -MY
lowess was old-fashioned a decade ago: use loess. And this Q was answered about a week ago, so use the archives. On Thu, 10 Apr 2003, Minghua Yao wrote:> I want to use lowess(x, y) where x and y are vectors of length of 4000+. In > fact, x and y are log of some vectors. So, some of the elements are NaN. > lowess() can not take away those elements then do the fitting. It will give > the error message and do nothing. > > 1. Can anybody tell me how to get rid of those NaN's and use lowess()? > 2. How to get the LOWESS fitting values for any elements in x?-- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595
"Statistical Models in S" edited by Chamber & Hastie (the "white book") has a chapter on loess. "Modern Applied statistics in S" by Venables & Ripley has some coverage as well. On top of that, you have the source code--- you can get as detailed as you wish with that! Andy> -----Original Message----- > From: Minghua Yao [mailto:myao at ou.edu] > Sent: Thursday, April 10, 2003 3:58 PM > To: Prof Brian Ripley > Cc: r-help at stat.math.ethz.ch > Subject: RE: [R] A Question on lowess() function > > > Thank you for your reply. > > I didn't find what I needed from the archieves. Maybe, I need > to figure out > how to search the archieves effectively. > > I used y<-x[!is.na(x)] to get rid of NA and NaN. But I don't > know how to get > rid of Inf. > > Also, is there more detailed info about loess() than help(loess)? > > Thanks. > > -MY > > -----Original Message----- > From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk] > Sent: Thursday, April 10, 2003 1:38 PM > To: Minghua Yao > Cc: r-help at stat.math.ethz.ch > Subject: Re: [R] A Question on lowess() function > > > lowess was old-fashioned a decade ago: use loess. > > And this Q was answered about a week ago, so use the archives. > > On Thu, 10 Apr 2003, Minghua Yao wrote: > > > I want to use lowess(x, y) where x and y are vectors of > length of 4000+. > In > > fact, x and y are log of some vectors. So, some of the > elements are NaN. > > lowess() can not take away those elements then do the > fitting. It will > give > > the error message and do nothing. > > > > 1. Can anybody tell me how to get rid of those NaN's and > use lowess()? > > 2. How to get the LOWESS fitting values for any elements in x? > > -- > Brian D. Ripley, ripley at stats.ox.ac.uk > Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ > University of Oxford, Tel: +44 1865 272861 (self) > 1 South Parks Road, +44 1865 272866 (PA) > Oxford OX1 3TG, UK Fax: +44 1865 272595 > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://www.stat.math.ethz.ch/mailman/listinfo/r-help >------------------------------------------------------------------------------
On the "See Also" section of the help page to loess(), there's a
link to
"predict.loess". On that help page, under "Arguments":
newdata an optional data frame specifying points at
which to do the predictions. If missing, the
original data points are used.
There, I've read the documentation for you.
Andy
> -----Original Message-----
> From: Minghua Yao [mailto:myao at ou.edu]
> Sent: Thursday, April 10, 2003 6:03 PM
> To: Prof Brian Ripley
> Cc: r-help at stat.math.ethz.ch
> Subject: RE: [R] A Question on lowess() function
>
>
> I still haven't found out from the mail archieves
>
> How to get the LOWESS or LOESS fitting values for any elements in x?
>
> Help please. Thanks.
>
> -MY
>
> -----Original Message-----
> From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
> Sent: Thursday, April 10, 2003 3:21 PM
> To: Minghua Yao
> Cc: r-help at stat.math.ethz.ch
> Subject: RE: [R] A Question on lowess() function
>
>
> On Thu, 10 Apr 2003, Minghua Yao wrote:
>
> > Thank you for your reply.
> >
> > I didn't find what I needed from the archieves. Maybe, I
> need to figure
> out
> > how to search the archieves effectively.
> >
> > I used y<-x[!is.na(x)] to get rid of NA and NaN. But I
> don't know how to
> get
> > rid of Inf.
>
> That's not what you asked for, and is.finite() will do that
> (if you apply
> it to x as well).
>
> > Also, is there more detailed info about loess() than help(loess)?
>
> Look at the na.action parameter ..., as well as the references.
>
> > Thanks.
> >
> > -MY
> >
> > -----Original Message-----
> > From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
> > Sent: Thursday, April 10, 2003 1:38 PM
> > To: Minghua Yao
> > Cc: r-help at stat.math.ethz.ch
> > Subject: Re: [R] A Question on lowess() function
> >
> >
> > lowess was old-fashioned a decade ago: use loess.
> >
> > And this Q was answered about a week ago, so use the archives.
> >
> > On Thu, 10 Apr 2003, Minghua Yao wrote:
> >
> > > I want to use lowess(x, y) where x and y are vectors of
> length of 4000+.
> > In
> > > fact, x and y are log of some vectors. So, some of the
> elements are NaN.
> > > lowess() can not take away those elements then do the
> fitting. It will
> > give
> > > the error message and do nothing.
> > >
> > > 1. Can anybody tell me how to get rid of those NaN's and
> use lowess()?
> > > 2. How to get the LOWESS fitting values for any elements in x?
>
>
>
> --
> Brian D. Ripley, ripley at stats.ox.ac.uk
> Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
> University of Oxford, Tel: +44 1865 272861 (self)
> 1 South Parks Road, +44 1865 272866 (PA)
> Oxford OX1 3TG, UK Fax: +44 1865 272595
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
>
------------------------------------------------------------------------------