Mercurial > hg > camir-aes2014

annotate toolboxes/MIRtoolbox1.3.2/somtoolbox/som_info.m @ 0:e9a9cd732c1e tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
rev   line source
wolffd@0 1 function som_info(sS,level)
wolffd@0 2
wolffd@0 3 %SOM_INFO Displays information on the given SOM Toolbox struct.
wolffd@0 4 %
wolffd@0 5 % som_info(sS,[level])
wolffd@0 6 %
wolffd@0 7 % som_info(sMap);
wolffd@0 8 % som_info(sData,3);
wolffd@0 9 % som_info({sMap,sData});
wolffd@0 10 % som_info(sMap.comp_norm{2});
wolffd@0 11 %
wolffd@0 12 % Input and output arguments ([]'s are optional):
wolffd@0 13 % sS (struct) SOM Toolbox struct
wolffd@0 14 % (cell array of structs) several structs in a cell array
wolffd@0 15 % [level] (scalar) detail level (1-4), default = 1
wolffd@0 16 %
wolffd@0 17 % For more help, try 'type som_info' or check out online documentation.
wolffd@0 18 % See also SOM_SET.
wolffd@0 19
wolffd@0 20 %%%%%%%%%%%%% DETAILED DESCRIPTION %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 21 %
wolffd@0 22 % som_info
wolffd@0 23 %
wolffd@0 24 % PURPOSE
wolffd@0 25 %
wolffd@0 26 % Display information of the given SOM Toolbox struct(s).
wolffd@0 27 %
wolffd@0 28 % SYNTAX
wolffd@0 29 %
wolffd@0 30 % som_info(sM)
wolffd@0 31 % som_info({sM,sD})
wolffd@0 32 % som_info(...,level)
wolffd@0 33 %
wolffd@0 34 % DESCRIPTION
wolffd@0 35 %
wolffd@0 36 % Display the contents of the given SOM Toolbox struct(s). Information
wolffd@0 37 % of several structs can be shown if the structs are given in a cell
wolffd@0 38 % array. The level of detail can be varied with the second argument.
wolffd@0 39 % The number of different levels varies between structs. For map and
wolffd@0 40 % data structs, not only the fields, but also some statistics of the
wolffd@0 41 % vectors ('.data' and '.codebook' fields) is displayed.
wolffd@0 42 %
wolffd@0 43 % map struct
wolffd@0 44 % level 1: name, dimension, topology, dimension, neigborhood function,
wolffd@0 45 % mask and training status
wolffd@0 46 % level 2: ..., training history
wolffd@0 47 % level 3: ..., vector component names, statistics and normalization status
wolffd@0 48 % level 4: ..., vector component normalizations
wolffd@0 49 %
wolffd@0 50 % data struct:
wolffd@0 51 % level 1: name, dimension, data set completeness statistics
wolffd@0 52 % level 2: ..., vector component names, statistics and normalization status
wolffd@0 53 % level 3: ..., vector component normalizations
wolffd@0 54 % level 4: ..., label statistics
wolffd@0 55 %
wolffd@0 56 % topology struct:
wolffd@0 57 % level 1: all fields
wolffd@0 58 %
wolffd@0 59 % train struct:
wolffd@0 60 % level 1: all fields
wolffd@0 61 %
wolffd@0 62 % normalization struct:
wolffd@0 63 % level 1: method, status
wolffd@0 64 % level 2: ..., parameters
wolffd@0 65 %
wolffd@0 66 % REQUIRED INPUT ARGUMENTS
wolffd@0 67 %
wolffd@0 68 % sS (struct) SOM Toolbox struct
wolffd@0 69 % (cell array of structs) several structs in a cell array
wolffd@0 70 %
wolffd@0 71 % OPTIONAL INPUT ARGUMENTS
wolffd@0 72 %
wolffd@0 73 % level (scalar) detail level (1-4), default = 1
wolffd@0 74 %
wolffd@0 75 % EXAMPLES
wolffd@0 76 %
wolffd@0 77 % som_info(sM)
wolffd@0 78 % som_info(sM,4)
wolffd@0 79 % som_info(sM.trainhist)
wolffd@0 80 % som_info(sM.comp_norm{3})
wolffd@0 81 %
wolffd@0 82 % SEE ALSO
wolffd@0 83 %
wolffd@0 84 % som_set Set fields and create SOM Toolbox structs.
wolffd@0 85
wolffd@0 86 % Copyright (c) 1999-2000 by the SOM toolbox programming team.
wolffd@0 87 % http://www.cis.hut.fi/projects/somtoolbox/
wolffd@0 88
wolffd@0 89 % Version 1.0beta ecco 110997
wolffd@0 90 % Version 2.0beta juuso 101199
wolffd@0 91
wolffd@0 92 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 93 %% check arguments
wolffd@0 94
wolffd@0 95 error(nargchk(1, 2, nargin)) % check no. of input args is correct
wolffd@0 96
wolffd@0 97 if ~isstruct(sS),
wolffd@0 98 if ~iscell(sS) | ~isstruct(sS{1}),
wolffd@0 99 error('Invalid first input argument.')
wolffd@0 100 end
wolffd@0 101 csS = sS;
wolffd@0 102 else
wolffd@0 103 l = length(sS);
wolffd@0 104 csS = cell(l,1);
wolffd@0 105 for i=1:l, csS{i} = sS(i); end
wolffd@0 106 end
wolffd@0 107
wolffd@0 108 if nargin<2 | isempty(level) | isnan(level), level = 1; end
wolffd@0 109
wolffd@0 110 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0 111 %% print struct information
wolffd@0 112
wolffd@0 113 for c=1:length(csS),
wolffd@0 114 sS = csS{c};
wolffd@0 115 fprintf(1,'\n');
wolffd@0 116
wolffd@0 117 switch sS.type,
wolffd@0 118 case 'som_map',
wolffd@0 119 mdim = length(sS.topol.msize);
wolffd@0 120 [munits dim] = size(sS.codebook);
wolffd@0 121 t = length(sS.trainhist);
wolffd@0 122 if t==0, st='uninitialized';
wolffd@0 123 elseif t==1, st = 'initialized';
wolffd@0 124 else st = sprintf('initialized, trained %d times',t-1);
wolffd@0 125 end
wolffd@0 126
wolffd@0 127 % level 1
wolffd@0 128 fprintf(1,'Struct type : %s\n', sS.type);
wolffd@0 129 fprintf(1,'Map name : %s\n', sS.name);
wolffd@0 130 fprintf(1,'Input dimension : %d\n', dim);
wolffd@0 131 fprintf(1,'Map grid size : ');
wolffd@0 132 for i = 1:mdim - 1, fprintf(1,'%d x ',sS.topol.msize(i)); end
wolffd@0 133 fprintf(1,'%d\n', sS.topol.msize(mdim));
wolffd@0 134 fprintf(1,'Lattice type (rect/hexa) : %s\n', sS.topol.lattice);
wolffd@0 135 fprintf(1,'Shape (sheet/cyl/toroid) : %s\n', sS.topol.shape);
wolffd@0 136 fprintf(1,'Neighborhood type : %s\n', sS.neigh);
wolffd@0 137 fprintf(1,'Mask : ');
wolffd@0 138 if dim,
wolffd@0 139 for i = 1:dim-1, fprintf(1,'%d ',sS.mask(i)); end;
wolffd@0 140 fprintf(1,'%d\n',sS.mask(dim));
wolffd@0 141 else fprintf(1,'\n');
wolffd@0 142 end
wolffd@0 143 fprintf(1,'Training status : %s\n', st);
wolffd@0 144
wolffd@0 145 % level 1,
wolffd@0 146 status = cell(dim,1);
wolffd@0 147 for i=1:dim,
wolffd@0 148 n = length(sS.comp_norm{i});
wolffd@0 149 if n,
wolffd@0 150 uninit = strcmp('uninit',{sS.comp_norm{i}.status});
wolffd@0 151 done = strcmp('done',{sS.comp_norm{i}.status});
wolffd@0 152 undone = strcmp('undone',{sS.comp_norm{i}.status});
wolffd@0 153 if sum(uninit)==n, status{i} = 'none';
wolffd@0 154 elseif sum(done)==n, status{i} = 'done';
wolffd@0 155 elseif sum(undone)==n, status{i} = 'undone';
wolffd@0 156 else status{i} = 'partial';
wolffd@0 157 end
wolffd@0 158 else status{i} = 'no normalization'; end
wolffd@0 159 end
wolffd@0 160 if level>1,
wolffd@0 161 fprintf(1,'\nVector components\n');
wolffd@0 162 M = sS.codebook;
wolffd@0 163 fprintf(1,' # name mask min mean max std normalization\n');
wolffd@0 164 fprintf(1,' --- ------------ ---- ------ ------ ------ ------ -------------\n');
wolffd@0 165 for i = 1:dim,
wolffd@0 166 fprintf(1,' %-3d %-12s %-4.2f %6.2g %6.2g %6.2g %6.2g %s\n', ...
wolffd@0 167 i,sS.comp_names{i}, sS.mask(i), ...
wolffd@0 168 min(M(:,i)),mean(M(:,i)),max(M(:,i)),std(M(:,i)),status{i});
wolffd@0 169 end
wolffd@0 170 end
wolffd@0 171
wolffd@0 172 % level 3
wolffd@0 173 if level>2,
wolffd@0 174 fprintf(1,'\nVector component normalizations\n');
wolffd@0 175 fprintf(1,' # name method (i=uninit,u=undone,d=done)\n');
wolffd@0 176 fprintf(1,' --- ------------ ---------------------------------------\n');
wolffd@0 177 for i=1:dim,
wolffd@0 178 fprintf(1,' %-3d %-12s ',i,sS.comp_names{i});
wolffd@0 179 n = length(sS.comp_norm{i});
wolffd@0 180 for j=1:n,
wolffd@0 181 m = sS.comp_norm{i}(j).method;
wolffd@0 182 s = sS.comp_norm{i}(j).status;
wolffd@0 183 if strcmp(s,'uninit'), c='i';
wolffd@0 184 elseif strcmp(s,'undone'), c='u';
wolffd@0 185 else c='d';
wolffd@0 186 end
wolffd@0 187 fprintf(1,'%s[%s] ',m,c);
wolffd@0 188 end
wolffd@0 189 fprintf(1,'\n');
wolffd@0 190 end
wolffd@0 191 end
wolffd@0 192
wolffd@0 193 % level 4
wolffd@0 194 if level>3,
wolffd@0 195 fprintf(1,'\nTraining history\n');
wolffd@0 196 fprintf(1,'Algorithm Data Trainlen Neigh.f. Radius Alpha (type) Date\n');
wolffd@0 197 fprintf(1,'--------- ------------- -------- -------- ---------- -------------- --------------------\n');
wolffd@0 198 for i=1:t,
wolffd@0 199 sT = sS.trainhist(i);
wolffd@0 200 fprintf(1,'%8s %13s %8d %8s %4.2f->%4.2f %5.3f (%6s) %s\n',...
wolffd@0 201 sT.algorithm,sT.data_name,sT.trainlen,...
wolffd@0 202 sT.neigh,sT.radius_ini,sT.radius_fin,sT.alpha_ini,sT.alpha_type,sT.time);
wolffd@0 203 %for j = 1:length(sT.mask)-1, fprintf(1,'%d ',sT.mask(j)); end;
wolffd@0 204 %if ~isempty(sT.mask), fprintf(1,'%d\n',sT.mask(end)); else fprintf(1,'\n'); end
wolffd@0 205 end
wolffd@0 206 end
wolffd@0 207
wolffd@0 208 case 'som_data',
wolffd@0 209
wolffd@0 210 [dlen dim] = size(sS.data);
wolffd@0 211 if dlen*dim
wolffd@0 212 if dim>1, ind = find(~isnan(sum(sS.data,2)));
wolffd@0 213 else ind = find(~isnan(sS.data));
wolffd@0 214 end
wolffd@0 215 else ind = []; end
wolffd@0 216 complete = size(sS.data(ind,:),1);
wolffd@0 217 partial = dlen - complete;
wolffd@0 218 values = prod(size(sS.data));
wolffd@0 219 missing = sum(sum(isnan(sS.data)));
wolffd@0 220
wolffd@0 221 % level 1
wolffd@0 222 fprintf(1,'Struct type : %s\n', sS.type);
wolffd@0 223 fprintf(1,'Data name : %s\n', sS.name);
wolffd@0 224 fprintf(1,'Vector dimension : %d\n', dim);
wolffd@0 225 fprintf(1,'Number of data vectors : %d\n', dlen);
wolffd@0 226 fprintf(1,'Complete data vectors : %d\n', complete);
wolffd@0 227 fprintf(1,'Partial data vectors : %d\n', partial);
wolffd@0 228 if values, r = floor(100 * (values - missing) / values); else r = 0; end
wolffd@0 229 fprintf(1,'Complete values : %d of %d (%d%%)\n', ...
wolffd@0 230 values-missing, values, r);
wolffd@0 231
wolffd@0 232 % level 2,
wolffd@0 233 status = cell(dim,1);
wolffd@0 234 for i=1:dim,
wolffd@0 235 n = length(sS.comp_norm{i});
wolffd@0 236 if n,
wolffd@0 237 uninit = strcmp('uninit',{sS.comp_norm{i}.status});
wolffd@0 238 done = strcmp('done',{sS.comp_norm{i}.status});
wolffd@0 239 undone = strcmp('undone',{sS.comp_norm{i}.status});
wolffd@0 240 if sum(uninit)==n, status{i} = 'none';
wolffd@0 241 elseif sum(done)==n, status{i} = 'done';
wolffd@0 242 elseif sum(undone)==n, status{i} = 'undone';
wolffd@0 243 else status{i} = 'partial';
wolffd@0 244 end
wolffd@0 245 else status{i} = 'no normalization'; end
wolffd@0 246 end
wolffd@0 247 if level>1,
wolffd@0 248 fprintf(1,'\nVector components\n');
wolffd@0 249 D = sS.data;
wolffd@0 250 fprintf(1,' # name min mean max std missing normalization\n');
wolffd@0 251 fprintf(1,' --- ------------ ------ ------ ------ ------ ----------- -------------\n');
wolffd@0 252 for i = 1:dim,
wolffd@0 253 known = find(~isnan(D(:,i)));
wolffd@0 254 miss = dlen-length(known);
wolffd@0 255 switch length(known),
wolffd@0 256 case 0, mi = NaN; me = NaN; ma = NaN; st = NaN;
wolffd@0 257 case 1, mi = D(known,i); me = mi; ma = mi; st = 0;
wolffd@0 258 otherwise,
wolffd@0 259 mi = min(D(known,i)); ma = max(D(known,i));
wolffd@0 260 me = mean(D(known,i)); st = std(D(known,i));
wolffd@0 261 end
wolffd@0 262 fprintf(1,' %-3d %-12s %6.2g %6.2g %6.2g %6.2g %5d (%2d%%) %s\n', ...
wolffd@0 263 i,sS.comp_names{i},mi,me,ma,st,miss,floor(100*miss/dlen),status{i});
wolffd@0 264 end
wolffd@0 265 end
wolffd@0 266
wolffd@0 267 % level 3
wolffd@0 268 if level>2,
wolffd@0 269 fprintf(1,'\nVector component normalizations\n');
wolffd@0 270 fprintf(1,' # name method (i=uninit,u=undone,d=done)\n');
wolffd@0 271 fprintf(1,' --- ------------ ---------------------------------------\n');
wolffd@0 272 for i=1:dim,
wolffd@0 273 fprintf(1,' %-3d %-12s ',i,sS.comp_names{i});
wolffd@0 274 n = length(sS.comp_norm{i});
wolffd@0 275 for j=1:n,
wolffd@0 276 m = sS.comp_norm{i}(j).method;
wolffd@0 277 s = sS.comp_norm{i}(j).status;
wolffd@0 278 if strcmp(s,'uninit'), c='i';
wolffd@0 279 elseif strcmp(s,'undone'), c='u';
wolffd@0 280 else c='d';
wolffd@0 281 end
wolffd@0 282 fprintf(1,'%s[%s] ',m,c);
wolffd@0 283 end
wolffd@0 284 fprintf(1,'\n');
wolffd@0 285 end
wolffd@0 286 end
wolffd@0 287
wolffd@0 288 % level 4
wolffd@0 289 if level>3,
wolffd@0 290 m = size(sS.labels,2);
wolffd@0 291 fprintf(1,'\nLabels\n');
wolffd@0 292 if isempty(sS.label_names),
wolffd@0 293 labs = {''}; freq = 0;
wolffd@0 294 for i=1:dlen*m,
wolffd@0 295 l = sS.labels{i};
wolffd@0 296 if isempty(l), freq(1) = freq(1)+1;
wolffd@0 297 else
wolffd@0 298 k = find(strcmp(labs,l));
wolffd@0 299 if isempty(k), labs{end+1} = l; freq(end+1) = 1;
wolffd@0 300 else freq(k)=freq(k)+1;
wolffd@0 301 end
wolffd@0 302 end
wolffd@0 303 end
wolffd@0 304 emp = freq(1);
wolffd@0 305 uni = length(freq)-1;
wolffd@0 306 if uni>0, tot = sum(freq(2:end)); else tot = 0; end
wolffd@0 307 fprintf(1,' Total: %d\n Empty: %d\n Unique: %d\n',tot,emp,uni);
wolffd@0 308 else
wolffd@0 309 for j=1:m,
wolffd@0 310 labs = {''}; freq = 0;
wolffd@0 311 for i=1:dlen,
wolffd@0 312 l = sS.labels{i,j};
wolffd@0 313 if isempty(l), freq(1) = freq(1)+1;
wolffd@0 314 else
wolffd@0 315 k = find(strcmp(labs,l));
wolffd@0 316 if isempty(k), labs{end+1} = l; freq(end+1) = 1;
wolffd@0 317 else freq(k)=freq(k)+1;
wolffd@0 318 end
wolffd@0 319 end
wolffd@0 320 end
wolffd@0 321 emp = freq(1);
wolffd@0 322 uni = length(freq)-1;
wolffd@0 323 if uni>0, tot = sum(freq(2:end)); else tot = 0; end
wolffd@0 324 fprintf(1,' [%s] Total / empty / unique: %d / %d / %d\n', ...
wolffd@0 325 sS.label_names{j},tot,emp,uni);
wolffd@0 326 end
wolffd@0 327 end
wolffd@0 328 end
wolffd@0 329
wolffd@0 330 case 'som_topol',
wolffd@0 331
wolffd@0 332 mdim = length(sS.msize);
wolffd@0 333
wolffd@0 334 % level 1
wolffd@0 335 fprintf(1,'Struct type : %s\n',sS.type);
wolffd@0 336 fprintf(1,'Map grid size : ');
wolffd@0 337 for i = 1:mdim - 1, fprintf(1,'%d x ',sS.msize(i)); end
wolffd@0 338 fprintf(1,'%d\n', sS.msize(mdim));
wolffd@0 339 fprintf(1,'Lattice type (rect/hexa) : %s\n', sS.lattice);
wolffd@0 340 fprintf(1,'Shape (sheet/cyl/toroid) : %s\n', sS.shape);
wolffd@0 341
wolffd@0 342 case 'som_train',
wolffd@0 343
wolffd@0 344 % level 1
wolffd@0 345 fprintf(1,'Struct type : %s\n',sS.type);
wolffd@0 346 fprintf(1,'Training algorithm : %s\n',sS.algorithm);
wolffd@0 347 fprintf(1,'Training data : %s\n',sS.data_name);
wolffd@0 348 fprintf(1,'Neighborhood function : %s\n',sS.neigh);
wolffd@0 349 fprintf(1,'Mask : ');
wolffd@0 350 dim = length(sS.mask);
wolffd@0 351 if dim,
wolffd@0 352 for i = 1:dim-1, fprintf(1,'%d ',sS.mask(i)); end;
wolffd@0 353 fprintf(1,'%d\n',sS.mask(end));
wolffd@0 354 else fprintf(1,'\n'); end
wolffd@0 355 fprintf(1,'Initial radius : %-6.1f\n',sS.radius_ini);
wolffd@0 356 fprintf(1,'Final radius : %-6.1f\n',sS.radius_fin);
wolffd@0 357 fprintf(1,'Initial learning rate (alpha) : %-6.1f\n',sS.alpha_ini);
wolffd@0 358 fprintf(1,'Alpha function type (linear/inv) : %s\n',sS.alpha_type);
wolffd@0 359 fprintf(1,'Training length : %d\n',sS.trainlen);
wolffd@0 360 fprintf(1,'When training was done : %s\n',sS.time);
wolffd@0 361
wolffd@0 362 case 'som_norm',
wolffd@0 363
wolffd@0 364 % level 1
wolffd@0 365 fprintf(1,'Struct type : %s\n',sS.type);
wolffd@0 366 fprintf(1,'Normalization method : %s\n',sS.method);
wolffd@0 367 fprintf(1,'Status : %s\n',sS.status);
wolffd@0 368
wolffd@0 369 % level 2
wolffd@0 370 if level>1,
wolffd@0 371 fprintf(1,'Parameters:\n');
wolffd@0 372 sS.params
wolffd@0 373 end
wolffd@0 374
wolffd@0 375 case 'som_grid',
wolffd@0 376
wolffd@0 377 % level 1
wolffd@0 378 fprintf(1,'Struct type : %s\n',sS.type);
wolffd@0 379 if ischar(sS.neigh),
wolffd@0 380 fprintf(1,'Connections : [%d %d], %s, %s\n',...
wolffd@0 381 sS.msize(1),sS.msize(2),sS.neigh,sS.shape);
wolffd@0 382 else
wolffd@0 383 fprintf(1,'Connections : [%d %d] %d lines\n',...
wolffd@0 384 sS.msize(1),sS.msize(2),sum(sS.neigh));
wolffd@0 385 end
wolffd@0 386 fprintf(1,'Line : %s\n',sS.line);
wolffd@0 387 if length(sS.marker)==1,
wolffd@0 388 fprintf(1,'Marker : %s\n',sS.marker);
wolffd@0 389 else
wolffd@0 390 fprintf(1,'Marker : varies\n');
wolffd@0 391 end
wolffd@0 392 fprintf(1,'Surf : ');
wolffd@0 393 if isempty(sS.surf), fprintf(1,'off\n'); else fprintf(1,'on\n'); end
wolffd@0 394 fprintf(1,'Labels : ');
wolffd@0 395 if isempty(sS.label), fprintf(1,'off\n');
wolffd@0 396 else fprintf(1,'on (%d)\n',sS.labelsize); end
wolffd@0 397
wolffd@0 398 end
wolffd@0 399
wolffd@0 400 fprintf(1,'\n');
wolffd@0 401 end
wolffd@0 402
wolffd@0 403 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%