The R Project SVN R

Rev

Rev 90081 | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 90081 Rev 90259
Line 120... Line 120...
120 
  \emph{inversions} or \emph{reversals} which are hard to interpret,
 120 
  \emph{inversions} or \emph{reversals} which are hard to interpret,
121 
  but note the trichotomies in \bibcitet{R:Legendre+Legendre:2012}.
 121 
  but note the trichotomies in \bibcitet{R:Legendre+Legendre:2012}.
122 
 
 122 
 
123 
  Two different algorithms are found in the literature for Ward
 123 
  Two different algorithms are found in the literature for Ward
124 
  clustering.  The one used by option \code{"ward.D"} (equivalent to the
 124 
  clustering.  The one used by option \code{"ward.D"} (equivalent to the
125 
  only Ward option \code{"ward"} in \R versions \eqn{\le}{<=} 3.0.3)
 125 
  only Ward option \code{"ward"} in \R versions \eqn{\le} 3.0.3)
126 
  \emph{does not} implement the clustering criterion of
 126 
  \emph{does not} implement the clustering criterion of
127 
  \bibcitet{R:Ward:1963}, whereas option \code{"ward.D2"} implements
 127 
  \bibcitet{R:Ward:1963}, whereas option \code{"ward.D2"} implements
128 
  that criterion \bibcitep{R:Murtagh+Legendre:2014}.  With the latter, the
 128 
  that criterion \bibcitep{R:Murtagh+Legendre:2014}.  With the latter, the
129 
  dissimilarities are \emph{squared} before cluster updating, and the
 129 
  dissimilarities are \emph{squared} before cluster updating, and the
130 
  clustering corresponds to \code{"ward.D"} with squared distances, see
 130 
  clustering corresponds to \code{"ward.D"} with squared distances, see