R/rcentroid.R

In dave: Functions for "Data Analysis in Vegetation Ecology"

rcentroid<- function(veg,membership.r,y,...){
    # computing a matrix of releve group centroids (rows) with all species(columns)
    # available as frequency table (species counts) or as probability (relative
    # frequency) 21. 3. 2012
    nrel <- length(veg[,1])
    nspec <- length(veg[1,])
    nele <- nrel*nspec                                 # no. of elements in the entire table
    xall <- as.numeric(as.matrix(t(veg),nrow=nele))    # entire matrix in one dimension
    rowgroups<- rep(seq(1,nspec,1),nrel)
    col <- c(rep(membership.r,nspec))
    col <- matrix(col,nrow=nrel,ncol=nspec)
    col <- t(col)
    colgroups <- matrix(col,nrow=nele)
    F <- tapply(sign(xall),list(rowgroups, colgroups),sum)      # contingency table, counts
    F<- t(F)
    colnames(F)<- colnames(veg)
    # relativized centroid table in P
    tt<-table(membership.r)
    S<-matrix(rep(tt,nspec),nrow=length(tt))
    P<- F/S
    colnames(P)<- colnames(veg)
    
    dgr<- (1 - cor(t(P^y)))/2
    dgr<- as.dist(dgr,upper=T,diag=T)
    
    # list output
    rcentroid<- list(nrelgroups=length(tt),nspec=nspec,freq.table=F,prob.table=P,dist.mat=dgr)
}