Skip to content

adehabitat in R

2 messages · carolang, Mathieu Basille

Original
1 day later
Mathieu Basille
# 
Dear Carolina,

I'm not sure what's your exact problem, since you don't provide any data, 
sample or code. If your matrix is such that rows define available units 
(usually, one row = one pixel of a raster map) over several variables (i.e. 
columns), you just need a vector of presence, with the number of 
observation associated to each available unit. This vector can be an 
additional column. Please have a look at the help of ?enfa, and 
specifically the example:

      data(lynxjura)

      map <- lynxjura$map

      ## We keep only "wild" indices.
      locs <- lynxjura$locs
      locs <- locs[slot(locs, "data")[,2]!="D",]

      hist(map, type = "l")
      ## The variable artif is far from symetric

      ## We perform a square root transformation
      ## of this variable
      ## We therefore normalize the variable 'artif'
      slot(map,"data")[,4] <- sqrt(slot(map,"data")[,4])
      hist(map, type = "l")

      ## We prepare the data for the ENFA
      tab <- slot(map, "data")
      pr <- slot(count.points(locs, map), "data")[,1]

You can then check what's in 'tab' and 'pr', which are the basic elements 
required for an ENFA:

 > head(tab)
      forets     hydro    routes     artif
1  5.154568 11.230920 11.622950 16.607230
2 11.643950 10.054470 11.389700  4.661158
3 16.126420  8.814120 10.972950  7.452453
4 18.757530  7.549436 10.417370  9.770423
5 19.779750  6.323049  9.784133 21.188813
6 22.390620  5.249640  9.140086 26.448665
 > head(pr)
[1] 0 0 0 0 0 0
 > table(pr)
pr
    0    1    2    3    4
8220  379   33    7    1

All that's needed then is to run a PCA on the variables measured on the 
available units (don't forget to remove the vector of presence if it's a 
column of your matrix), and use the result of this PCA together with the 
vector of presence:

      ## We then perform the PCA before the ENFA
      pc <- dudi.pca(tab, scannf = FALSE)

      (enfa1 <- enfa(pc, pr,
                     scannf = FALSE))

Hope this helps,
Mathieu.


Le 03/07/2012 10:36, carolang a ?crit :