R help - Oct 2008 - predicting from a local regression and plotting in lattice

Hi R community,

I'm running R 2.7.2 on Windows XP SP2.

I'm trying to (1)  plot loess lines for each of my groupings using the same
color for each group; (2) plot loess predicted values.

The first part is easy:
 data1 <-
data.frame(Names=c(rep("Jon",9),rep("Karl",9)),Measurements=c(2,4,16,25,36,49,64,81,100,1,2,5,12,17,21,45,54,67),PlotAt=c(1:9,1:9))

 data2 <-
data.frame(Names=c(rep("Jonah",9),rep("Beth",9)),Measurements=c(1:5,1,1,1,1,rep(20,9)),PlotAt=c(1:9,1:9))

xyplot(Measurements ~ PlotAt, data = data1, groups=Names,

 panel = function(...) {
    panel.superpose(...,col="gray",lty=1,lwd=2);
    },
  panel.groups = function(...) panel.loess(...)
  )

My question is, how do I plot predicted values from loess to extend the
lines beyond the range of x? Let's say I want all lines to stop at x = 15.
Is there an easy way to do this?


Thanks for any help,
-AK

	[[alternative HTML version deleted]]

Alex Karner <aakarner <at> ucdavis.edu> writes:
> I'm trying to (1)  plot loess lines for each of my groupings using the
same
> color for each group; (2) plot loess predicted values.
> 
> The first part is easy:
.. Example removed... Thanks, it was a good example of what you wanted!
> My question is, how do I plot predicted values from loess to extend the
> lines beyond the range of x? Let's say I want all lines to stop at x =
15.
loess is "local polynomial regression fitting". It is good to guide
your eyes
through confusing data points, but not at all suitable for extrapolation
(that's
why it is good locally). I would say, it is not even good for interpolation, but
that a matter of discussion.

I doubt that an extrapolation of a factor 2 beyond the initial range is possible
at all without a VERY good model of your data. But in each case, you need a
model, for example from earlier longer records. How does the poor function know
that your data do not bend down logistic-like after you last value?


Dieter

Thanks for your response, Dieter.

I realize these limitations. However, I know that my actual dataset is
reasonably well behaved in the range I want to predict, and I'm not using
the predicted values for any further analysis, only for schematic purposes
in the plot.

I'm still curious if this type of extension of a loess line is possible,
notwithstanding its statistical shortcomings.


Cheers,
Alex.
>loess is "local polynomial regression fitting". It is good to
guide your
eyes
>through confusing data points, but not at all suitable for extrapolation
(that's
>why it is good locally). I would say, it is not even good for
interpolation, but
>that a matter of discussion.
>I doubt that an extrapolation of a factor 2 beyond the initial range is
possible
>at all without a VERY good model of your data. But in each case, you need a
>model, for example from earlier longer records. How does the poor function
know
>that your data do not bend down logistic-like after you last value?
>
>
>Dieter
Alex Karner <aakarner <at> ucdavis.edu> writes:
> I'm trying to (1)  plot loess lines for each of my groupings using the
same
> color for each group; (2) plot loess predicted values.
>
> The first part is easy:
.. Example removed... Thanks, it was a good example of what you wanted!
> My question is, how do I plot predicted values from loess to extend the
> lines beyond the range of x? Let's say I want all lines to stop at x =
15.
	[[alternative HTML version deleted]]

locfit can extrapolate:

library(locfit)
X <- seq(min(time(Nile)), max(time(Nile))+50)

plot(range(X), range(Nile), type = "n")
lines(Nile)

fit <- locfit(Nile ~ time(Nile))
lines(X, predict(fit, new = X), col = "red")


On Fri, Oct 10, 2008 at 4:42 PM, Alex Karner <aakarner at ucdavis.edu>
wrote:
> Hi R community,
>
> I'm running R 2.7.2 on Windows XP SP2.
>
> I'm trying to (1)  plot loess lines for each of my groupings using the
same
> color for each group; (2) plot loess predicted values.
>
> The first part is easy:
>  data1 <-
>
data.frame(Names=c(rep("Jon",9),rep("Karl",9)),Measurements=c(2,4,16,25,36,49,64,81,100,1,2,5,12,17,21,45,54,67),PlotAt=c(1:9,1:9))
>
>  data2 <-
>
data.frame(Names=c(rep("Jonah",9),rep("Beth",9)),Measurements=c(1:5,1,1,1,1,rep(20,9)),PlotAt=c(1:9,1:9))
>
> xyplot(Measurements ~ PlotAt, data = data1, groups=Names,
>
>  panel = function(...) {
>    panel.superpose(...,col="gray",lty=1,lwd=2);
>    },
>  panel.groups = function(...) panel.loess(...)
>  )
>
> My question is, how do I plot predicted values from loess to extend the
> lines beyond the range of x? Let's say I want all lines to stop at x =
15.
> Is there an easy way to do this?
>
>
> Thanks for any help,
> -AK
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>