Christoph Budziszewski
starting som prediction fine-tuned class-performance visualisation
Christoph Budziszewski commited 4dbef18 at 2009-01-21 16:34:25
function [sC,old2new,newi] = som_clset(sC,action,par1,par2)
% SOM_CLSET Create and/or set values in the som_clustering struct.
%
% first argument
% sC (struct) a som_clustering struct
% Z (matrix) size nb-1 x 3, as given by LINKAGE function
% base (vector) size dlen x 1, a partitioning of the data
%
% actions
% 'remove' removes the indicated clusters (par1: vector)
% 'add' add a cluster by making a combination of the indicated
% clusters (par1: vector)
% %'move' moves a child cluster (par1: scalar) from a parent to another
% % (par2: vector 1 x 2)
% 'merge' like 'add', followed by removing the indicated clusters (par1: vector)
% %'split' the indicated cluster (par1: scalar) is partitioned into indicated
% % parts (par2: vector), which are then added, while the indicated cluster
% % (par1) is removed
% 'coord' sets the coordinates of base clusters (par1: matrix nb x *), and
% recalculates coordinates of the derived clusters (by averaging base cluster
% coordinates)
% 'color' sets the colors of base clusters (par1: matrix nb x 3), and recalculates
% colors of the derived clusters (as averages of base cluster colors)
%
% sC
% .type (string) 'som_clustering'
% .name (string) Identifier for the clustering.
% .nb (scalar) Number of base clusters in the clustering.
% .base (vector) Size dlen x 1, the basic groups of data
% forming the base clusters, e.g. as a result
% of partitive clustering. Allowed values are
% 1:nb indicating the base cluster
% to which the data belongs to.
% NaN indicating that the data has
% been ignored in the clustering
% .nc (scalar) Number of clusters in the clustering (nb + derived clusters).
% .children (cellarray) size nc x 1, each cell gives the list of indeces
% of child clusters for the cluster
% .parent (vector) size nc x 1, the index of parent of each cluster
% (or zero if the cluster does not have a parent)
% .coord (matrix) size nc x *, visualization coordinates for each cluster
% By default the coordinates are set so that
% the base clusters are ordered on a line, and the
% position of each combined cluster is average of
% the base clusters that constitute it.
% .color (matrix) size nc x 3, color for each cluster.
% By default the colors are set so that the
% base clusters are ordered on a line,
% and then colors are assigned from the 'hsv'
% colormap to the base clusters. The color
 
הההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההההה
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX