Computes the two first discriminant coordinates (canonical coordinates) directly from a tclust
object to obtain a graphical representations of cluster solutions in higher dimensional (p > 2) cases.
discr_coords(x, equal.weights)
A tclust
object.
A logical value, controlling whether the clusters should be considered as equal-sized (TRUE
) when combining their covariance structures, or if their actual size shall be considered (FALSE
). By default value x$par$equal.weights
is assumed
A two-dimensional matrix, containing the canonical coordinates of all observations given by the tclust
-object.
The functionality of discr_coords
is directly derived from discrcoord
as implemented in the package "fpc" by Christian Hennig. It has been adopted in order to directly use the covariance information contained in the tclust
-object. The function fails, if "store.x = FALSE
" is specified in tclust
, because the original data matrix is required here.
Hennig, C. and Christlieb, N. (2002), "Validating visual clusters in large datasets: fixed point clusters of spectral features.", Computational Statistics and Data Analysis Vol.40, 723-739.