[R-sig-Geo] Using a mask

[Roger Bivand]

On Fri, 27 Feb 2009, Ned Horning wrote:

> Is there an easy way to mask (or flag) data values in R so they do not get 
> processed. I am using predict.randomForest on a satellite image and there are 
> a lot (~30%) of pixels with a value of "0" that I'd prefer not to process 
> with predict.randomForest to save time. I was thinking it might be possible 
> to create a data frame with valid data for each line read from the input 
> multi-band image and then reconstruct the line before writing it out but I 
> think the overhead would be very high. At least using the brute force methods 
> like rbind.

It looks as though setting the values to NA will work. In the 
predict.randomForest example, do:

df <- iris[ind == 2,]
is.na(df[1:10, 1:2]) <- TRUE
predict(iris.rf, df)

So setting the 0 values to NA might get you the results you want, 
maintaining the relative order of the observations.

Roger

>
> All the best,
>
> Ned
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no