Skip to content

kriging output

5 messages · Dave Depew, Edzer Pebesma, Roger Bivand

Original
Dave Depew
# 
Hi all,
I've got a question regarding kriging outputs. I have an interpolated 
dataset which due to the nugget effect contains some negative values as 
the predictions. I would like to truncate these @ "0", rather than 
having them as a negative prediction.
I've tried something similar with the meuse data set...

data(meuse.grid) 
coordinates(meuse.grid) = c("x", "y") 
gridded(meuse.grid) <- TRUE
meuse.grid[["idist"]] = 1 - meuse.grid[["dist"]] 

Somewhat arbitrary, but essentially, I'd like to set all "dist" data less than 0.5 to "0", and have the rest remain as they are...


meuse.grid[["class"]] = for(i in 1:length(meuse.grid[["dist"]])){
if (meuse.grid[["dist"]]<0.5) {meuse.grid[["class"]]==0} else
{meuse.grid[["class"]]=meuse.grid[["dist"]]}
}

All I've managed to get is the new attribute "class" all filled with 0.

Any insight would be appreciated.
Edzer Pebesma
#
Dave Depew wrote:
I guess you meant an = rather than a == here
Try:

meuse.grid$class = meuse.grid$dist
meuse.grid$class[meuse.grid$dist < .5] = 0
plot(class~dist,meuse.grid)
--
Edzer
Dave Depew
# 
Thanks,
This worked.

I'm still confused why the if else statement didn't work...
If one wanted to do conditional arithmetic would a for statement bee needed?


e.g.
 meuse.grid[["class"]] = for(i in 1:length(meuse.grid[["dist"]])){
if (meuse.grid[["dist"]]<0.5) 
{meuse.grid[["class"]]=10*meuse.grid[["dist]]} else
{meuse.grid[["class"]]=100* meuse.grid[["dist"]]}
 }
Edzer Pebesma wrote:
Edzer Pebesma
#
Dave Depew wrote:
I'd say
meuse.grid$class = ifelse(meuse.grid$dist < .5, 10, 100) * meuse.grid$dist

assigning the result from a for statement also does not yield a beauty 
price if you ask me, but that might be my  backgrounds in C, long ago.
--
Edzer
Roger Bivand
#
On Mon, 5 May 2008, Edzer Pebesma wrote:

            

      
      
      
    

        
Right. ifelse() is vectorised, but if() else is not, so the vector 
condition in ifelse() splits out nicely, but if() else is flow control on 
a scalar condition. The above for loop looks redundant anyway (no use of 
index), but the if() else construction wouldn't work either - as you 
found out earlier.

For things like this, Braun & Murdoch is good (and from Canada!); if you 
read German, Uwe Ligges' book is recommended. It's also in the online 
Introduction to R, section 9.2.1, which explains the previous paragraph 
much more clearly than I can.

Roger