Daniel@0: function sD = som_modify_dataset(sD,action,varargin)
Daniel@0: 
Daniel@0: %SOM_MODIFY_DATASET Add or remove components or samples to/from a data struct.
Daniel@0: %
Daniel@0: % sD = som_modify_dataset(sD, 'addcomp', D, [indsto], [cnames})
Daniel@0: % sD = som_modify_dataset(sD, 'addsamp', D, [indsto], ['norm'])
Daniel@0: % sD = som_modify_dataset(sD, 'removecomp', [inds])
Daniel@0: % sD = som_modify_dataset(sD, 'removesamp', inds)
Daniel@0: % sD = som_modify_dataset(sD, 'extractcomp', [inds])
Daniel@0: % sD = som_modify_dataset(sD, 'extractsamp', inds)
Daniel@0: % sD = som_modify_dataset(sD, 'movecomp', inds, indsto)
Daniel@0: % sD = som_modify_dataset(sD, 'movesamp', inds, indsto)
Daniel@0: %
Daniel@0: %  Input and output arguments ([]'s are optional)
Daniel@0: %   sD      (struct) data struct
Daniel@0: %   action  (string) 'addcomp', 'addsamp', 'removecomp', 'removesamp', 
Daniel@0: %                    'extractcomp', 'extractsamp', 'movecomp', or 'movesamp' 
Daniel@0: %
Daniel@0: %   other input arguments depend on the action
Daniel@0: %
Daniel@0: %   'addcomp': 
Daniel@0: %   D        (matrix) data matrix, size [dlen x d]
Daniel@0: %            (struct) data struct, size of .data field [dlen x d]
Daniel@0: %   [indsto] (vector) new indeces of the components, length=d
Daniel@0: %   [cnames] (cellstr) of size d x 1, the component names
Daniel@0: %
Daniel@0: %   'addsamp': 
Daniel@0: %   D        (matrix) data matrix, size [n x dim] 
Daniel@0: %   [indsto] (vector) new indeces of the samples, length=n
Daniel@0: %   ['norm'] (string) specified if the normalization procedure
Daniel@0: %                     should be applied to the new samples
Daniel@0: %
Daniel@0: %   'removecomp', 'extractcomp': 
Daniel@0: %   [inds]   (vector) indeces of the components to be removed/extracted. 
Daniel@0: %                     If not given, a prompt will appear from which the
Daniel@0: %                     user can select the appropriate components.
Daniel@0: %
Daniel@0: %   'removesamp', 'extractsamp': 
Daniel@0: %   inds     (vector) indeces of the samples to be removed/extracted
Daniel@0: %
Daniel@0: %   'movecomp', 'movesamp': 
Daniel@0: %   inds     (vector) indeces of the components/samples to be moved
Daniel@0: %   indsto   (vector) new indeces of the components/samples 
Daniel@0: %
Daniel@0: % See also SOM_DATA_STRUCT.
Daniel@0: 
Daniel@0: % Copyright (c) 2000 by Juha Vesanto
Daniel@0: % Contributed to SOM Toolbox on June 16th, 2000 by Juha Vesanto
Daniel@0: % http://www.cis.hut.fi/projects/somtoolbox/
Daniel@0:  
Daniel@0: % Version 2.0beta juuso 200400 160600 280800
Daniel@0: 
Daniel@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Daniel@0: %% input arguments
Daniel@0: 
Daniel@0: [dlen dim] = size(sD.data);
Daniel@0: 
Daniel@0: switch action, 
Daniel@0:   case 'addcomp',  
Daniel@0:     D = varargin{1};     
Daniel@0:     if isstruct(D), [n d] = size(D.data); else [n d] = size(D); end
Daniel@0:     if n ~= dlen, error('The number of samples in the data struct and new data should match.'); end
Daniel@0:     indsto = []; 
Daniel@0:     cnames = []; 
Daniel@0:     for i=2:length(varargin), 
Daniel@0:       if isnumeric(varargin{i}), 
Daniel@0: 	indsto = varargin{i}; 
Daniel@0: 	if length(indsto) ~= d, 
Daniel@0: 	  error('The number of indeces should match the number of new components'); 
Daniel@0: 	end
Daniel@0:       else
Daniel@0: 	if ischar(varargin{i}), cnames = cellstr(varargin{i}); 
Daniel@0: 	elseif iscellstr(varargin{i}), cnames = varargin{i}; 
Daniel@0: 	else
Daniel@0: 	  error(['[som_modify_dataset] Unrecognized argument #' num2str(i+1)]); 
Daniel@0: 	end	
Daniel@0: 	if length(cnames) ~= d, 
Daniel@0: 	  error('The number of component names should match the number of new components'); 
Daniel@0: 	end
Daniel@0:       end
Daniel@0:     end
Daniel@0:  case 'addsamp',
Daniel@0:   D = varargin{1};
Daniel@0:   if isstruct(D), 
Daniel@0:     lab = D.labels; 
Daniel@0:     if isfield(D,'data'), D = D.data; else D = D.codebook; end
Daniel@0:   else lab = []; 
Daniel@0:   end
Daniel@0:   [n d] = size(D); 
Daniel@0:   if d ~= dim, 
Daniel@0:     error(['The dimensions of the old and new data sets should match.']); 
Daniel@0:   end
Daniel@0:   norm = 0; 
Daniel@0:   indsto = []; 
Daniel@0:   for i=2:length(varargin),
Daniel@0:     if ischar(varargin{i}) & strcmp(varargin{i},'norm'), norm = 1; 
Daniel@0:     elseif isnumeric(varargin{i}), 
Daniel@0:       indsto = varargin{i}; 
Daniel@0:       if length(indsto) ~= n, 
Daniel@0: 	error(['The number of new indeces should match the number of new' ...
Daniel@0: 	       ' samples']); 
Daniel@0:       end
Daniel@0:     else
Daniel@0:       warning(['[som_modify_dataset] Ignoring unrecognized argument #', ...
Daniel@0: 	       num2str(i+2)]);
Daniel@0:     end
Daniel@0:   end
Daniel@0:  case 'removecomp',
Daniel@0:   if length(varargin)>0, 
Daniel@0:     inds = varargin{1};
Daniel@0:   else
Daniel@0:     [inds, ok] = listdlg('ListString',sD.comp_names, 'PromptString', ...
Daniel@0:                          'Components', 'Name', 'Remove components', 'uh', 25);
Daniel@0:     if ~ok, return; end
Daniel@0:   end
Daniel@0:   if min(inds)<1 | max(inds)>dim, 
Daniel@0:     error('The component indeces must be within [1,dim]'); 
Daniel@0:   end
Daniel@0:  case 'removesamp',
Daniel@0:   inds = varargin{1};
Daniel@0:   if min(inds)<1 | max(inds)>dlen, 
Daniel@0:     error('The sample indeces must be within [1,dlen]'); 
Daniel@0:   end
Daniel@0:  case 'extractcomp',
Daniel@0:   if length(varargin)>0, 
Daniel@0:     inds = varargin{1};
Daniel@0:   else
Daniel@0:     [inds, ok] = listdlg('ListString',sD.comp_names, 'PromptString',... 
Daniel@0:                          'Components', 'Name', 'Extract components', 'uh', 25);
Daniel@0:     if ~ok, return; end
Daniel@0:   end
Daniel@0:   if min(inds)<1 | max(inds)>dim, 
Daniel@0:     error('The component indeces must be within [1,dim]'); 
Daniel@0:   end
Daniel@0:  case 'extractsamp',
Daniel@0:   inds = varargin{1};
Daniel@0:   if min(inds)<1 | max(inds)>dlen, 
Daniel@0:     error('The sample indeces must be within [1,dlen]'); 
Daniel@0:   end
Daniel@0:  case 'movecomp',
Daniel@0:   inds = varargin{1};
Daniel@0:   indsto = varargin{2};
Daniel@0:   if min(inds)<1 | max(inds)>dim | min(indsto)<1 | max(indsto)>dim, 
Daniel@0:     error('The component indeces must be within [1,dim]'); 
Daniel@0:   end
Daniel@0:  case 'movesamp',
Daniel@0:   inds = varargin{1};
Daniel@0:   indsto = varargin{2};
Daniel@0:   if min(inds)<1 | max(inds)>dlen | min(indsto)<1 | max(indsto)>dlen, 
Daniel@0:     error('The sample indeces must be within [1,dlen]'); 
Daniel@0:   end
Daniel@0:  otherwise, 
Daniel@0:   error('Unrecognized action mode');
Daniel@0: end   
Daniel@0: 
Daniel@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Daniel@0: %% action
Daniel@0: 
Daniel@0: switch action, 
Daniel@0:   case 'addcomp', 
Daniel@0:     if isstruct(D),       
Daniel@0:       sD.data = [sD.data, D.data]; 
Daniel@0:       sD.comp_names(dim+[1:d]) = D.comp_names; 
Daniel@0:       sD.comp_norm(dim+[1:d]) = D.comp_norm; 
Daniel@0:       sD = som_label(sD,'add',1:dlen,D.labels);
Daniel@0:     else
Daniel@0:       sD.data = [sD.data, D];
Daniel@0:       if isempty(cnames), 
Daniel@0: 	for i=1:d, sD.comp_names(dim+i) = {sprintf('Variable%d',i+dim)}; end
Daniel@0:       else
Daniel@0: 	sD.comp_names(dim+[1:d]) = cnames; 
Daniel@0:       end
Daniel@0:       for i=1:d, sD.comp_norm(dim+i) = {[]}; end
Daniel@0:     end
Daniel@0:     if ~isempty(indsto), 
Daniel@0:       sD = som_modify_dataset(sD,'movecomp',dim+[1:d],indsto);
Daniel@0:     end
Daniel@0: 
Daniel@0:   case 'addsamp', 
Daniel@0:     if norm, D = som_normalize(D,sD); end
Daniel@0:     sD.data = [sD.data; D]; 
Daniel@0:     nl = size(sD.labels,2); 
Daniel@0:     sD.labels(dlen+[1:n],1:nl) = {''};
Daniel@0:     if ~isempty(lab), 
Daniel@0:       nl2 = size(lab,2); 
Daniel@0:       if nl2>nl, sD.labels(1:dlen,nl+[1:(nl2-nl)]) = {''}; end
Daniel@0:       sD.labels(dlen+[1:n],1:nl2) = lab; 
Daniel@0:     end
Daniel@0:     if ~isempty(indsto), 
Daniel@0:       sD = som_modify_dataset(sD,'movesamp',dlen+[1:n],indsto);
Daniel@0:     end
Daniel@0: 
Daniel@0:   case 'removecomp', 
Daniel@0:     includeinds = 1:dim; 
Daniel@0:     includeinds(inds) = 0; 
Daniel@0:     sD = som_modify_dataset(sD,'extractcomp',find(includeinds));
Daniel@0: 
Daniel@0:   case 'removesamp', 
Daniel@0:     includeinds = 1:dlen; 
Daniel@0:     includeinds(inds) = 0; 
Daniel@0:     sD = som_modify_dataset(sD,'extractsamp',find(includeinds));
Daniel@0: 
Daniel@0:   case 'extractcomp', 
Daniel@0:     sD.data = sD.data(:,inds);
Daniel@0:     sD.comp_names = sD.comp_names(inds);
Daniel@0:     sD.comp_norm = sD.comp_norm(inds);    
Daniel@0: 
Daniel@0:   case 'extractsamp', 
Daniel@0:     sD.data = sD.data(inds,:);
Daniel@0:     sD.labels = sD.labels(inds,:);
Daniel@0: 
Daniel@0:   case 'movecomp', 
Daniel@0:     [indsto,order] = sort(indsto);
Daniel@0:     inds = inds(order);
Daniel@0:     oldinds = 1:dim; 
Daniel@0:     oldinds(inds) = 0; 
Daniel@0:     newinds = oldinds(oldinds>0);
Daniel@0:     for i=1:length(indsto),
Daniel@0:       ifrom = inds(i); ito = indsto(i);
Daniel@0:       if ito==1, newinds = [ifrom, newinds];
Daniel@0:       else newinds = [newinds(1:ito-1), ifrom, newinds(ito:end)];
Daniel@0:       end
Daniel@0:     end
Daniel@0:     sD.data = sD.data(:,newinds); 
Daniel@0:     sD.comp_names = sD.comp_names(:,newinds); 
Daniel@0:     sD.comp_norm = sD.comp_norm(:,newinds); 
Daniel@0: 
Daniel@0:   case 'movesamp', 
Daniel@0:     [indsto,order] = sort(indsto);
Daniel@0:     inds = inds(order);
Daniel@0:     oldinds = 1:dim; 
Daniel@0:     oldinds(inds) = 0; 
Daniel@0:     newinds = oldinds(oldinds>0);
Daniel@0:     for i=1:length(indsto),
Daniel@0:       ifrom = inds(i); ito = indsto(i);
Daniel@0:       if ito==1, newinds = [ifrom, newinds];
Daniel@0:       else newinds = [newinds(1:ito-1), ifrom, newinds(ito:end)];
Daniel@0:       end
Daniel@0:     end
Daniel@0:     sD.data = sD.data(newinds,:);
Daniel@0:     sD.labels = sD.labels(newinds,:);
Daniel@0: 
Daniel@0: end
Daniel@0: 
Daniel@0: %som_set(sD);
Daniel@0: 
Daniel@0: