[R] Smooth line in graph

[Greg Snow]

I don't know what SigmaPlot and Excel are doing for you, but I would guess that they are not doing cubic splines (as a general rule, when R and Excel differ, it is safest to assume that R is not the one doing something wrong)

Often differences between packages are due to differences in assumptions or model specifications.  I don't know SigmaPlot to be able to say what it is doing and a quick search of the excel help for 'spline' was not enlightening.

Perhaps what you are looking for is not a cubic spline.  Another option is an xspline.  These are implemented in the grid package.  Try:

library(grid)
library(lattice)

x<-c(-45,67,131,259,347)
y <- c(0.31, 0.45, 0.84, 0.43, 0.25)

s <- seq(-1,1, .25)

tmp.df <- data.frame(x=rep(x,9), y=rep(y,9), s=rep(s, each=5))

xyplot(y~x|factor(s), data=tmp.df, tmp.s=tmp.df$s, panel=function(x,y,subscripts,tmp.s,...){
 grid.points(x,y)
 tmp2.s <- tmp.s[subscripts]
print(tmp2.s)
 grid.xspline(x,y, default.units='native', shape=tmp2.s)
 })

Hope this helps,

-- 
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at intermountainmail.org
(801) 408-8111
 
 

> -----Original Message-----
> From: Nestor Fernandez [mailto:nestor at ual.es] 
> Sent: Thursday, September 20, 2007 4:02 AM
> To: r-help at stat.math.ethz.ch
> Cc: Greg Snow
> Subject: Re: [R] Smooth line in graph
> 
> Sorry, I answered too quickly.
> It worked with the "simplified" example I provided but not 
> with non-regular intervals in x:
> 
> x<-c(-45,67,131,259,347)
> y <- c(0.31, 0.45, 0.84, 0.43, 0.25)
> 
> plot(x,y)
> lines(spline(x,y, method='n', n=250))
> #or:
> lines(predict(interpSpline(x, y)))
> 
> Produce the same decrease between first two points and the 
> shape is quite different to that produced by sigmaplot -I 
> need comparable figures. Changing "method" and "n" arguments 
> did not help.
> 
> Sorry for bothering. Any other suggestion?
> 
> 
> 
> 
> 
> 
> Greg Snow escribió:
> > Try:
> >
> >   
> >> lines(spline(x,y, method='n', n=250))
> >>     
> >
> >   
> 
>