wolffd@0: function out = test_generic_display_results_absviolated(file)
wolffd@0: % out = test_generic_display_results_absviolated(file) 
wolffd@0: %
wolffd@0: % displays the finalresults mat file and enables 
wolffd@0: % further analysis and duiagnostics of the individual runs
wolffd@0: 
wolffd@0: global db_MTTAudioFeatureSlaney08;
wolffd@0: global db_magnamixedfeat_genrebasicsm;
wolffd@0: global db_MTTMixedFeatureGenreBasicSmPCA;
wolffd@0: 
wolffd@0: 
wolffd@0: global comparison;
wolffd@0: global comparison_ids;
wolffd@0:         
wolffd@0: if nargin < 1 || isempty(file) || isnumeric(file)
wolffd@0:     u = dir();
wolffd@0:     u = {u.name};
wolffd@0:     [idx, strpos] = substrcellfind(u, '_finalresults.mat', 1);
wolffd@0:     
wolffd@0:     if exist('file','var') && isnumeric(file)
wolffd@0:         file = u{idx(file)};
wolffd@0:     else
wolffd@0:         file = u{idx(1)};
wolffd@0:     end
wolffd@0: end
wolffd@0: 
wolffd@0: load(file);
wolffd@0: 
wolffd@0: % ---
wolffd@0: % % get statistics for feature parameters
wolffd@0: % Visualise the accuracy and variance
wolffd@0: % ---
wolffd@0: if isfield(out, 'inctrain')
wolffd@0:     for i = 1:numel(out)
wolffd@0:         
wolffd@0:         % ---
wolffd@0:         % get training and test sample sizes
wolffd@0:         % ---
wolffd@0:         nData = [];
wolffd@0:         n_train_data = zeros(1,numel(out(i).inctrain.dataPartition));
wolffd@0:         n_test_data = zeros(1,numel(out(i).inctrain.dataPartition));
wolffd@0:         for j = 1:numel(out(i).inctrain.dataPartition)
wolffd@0:             n_train_data(j) = mean(out(i).inctrain.dataPartition(j).TrainSize);
wolffd@0:             n_test_data(j) = mean(out(i).inctrain.dataPartition(j).TestSize);
wolffd@0:         end
wolffd@0:         
wolffd@0:         % ---
wolffd@0:         % get lost percentages
wolffd@0:         % ---
wolffd@0:         mean_lost_test = 1 - [out(i).inctrain.mean_ok_test];
wolffd@0:         mean_lost_test = mean_lost_test(1,:).* n_test_data;
wolffd@0:         
wolffd@0:         var_ok_test = sqrt([out(i).inctrain.var_ok_test]);
wolffd@0:         
wolffd@0:         mean_lost_train = 1 - [out(i).inctrain.mean_ok_train];
wolffd@0:         mean_lost_train = mean_lost_train(1,:).* n_train_data;
wolffd@0:         
wolffd@0:         % plot test results
wolffd@0:         figure;
wolffd@0:         subplot(2,1,1)
wolffd@0:         plot(n_train_data, mean_lost_test,'r');
wolffd@0:         hold;
wolffd@0:         % plot training results
wolffd@0:         plot(n_train_data, mean_lost_train,'m');
wolffd@0:         
wolffd@0:         xlabel ('# training constraints');
wolffd@0:         ylabel ('# constraints violated');
wolffd@0:         legend ('test','training');
wolffd@0:         
wolffd@0:         plot(n_train_data, mean_lost_test + var_ok_test(1,:).* n_test_data,'r:');
wolffd@0:         plot(n_train_data, mean_lost_test - (var_ok_test(1,:).* n_test_data),'r:');
wolffd@0:         
wolffd@0:         % ---
wolffd@0:         % get percentage of unknown data examples learnt
wolffd@0:         % ---
wolffd@0:         lost_test_not_in_train = mean_lost_test - mean_lost_train;
wolffd@0:         ntest_not_in_train = n_test_data - n_train_data;
wolffd@0:         
wolffd@0:         lost_test_not_in_train = lost_test_not_in_train ./ ntest_not_in_train;
wolffd@0:     
wolffd@0:         lost_test_not_in_train(isnan(lost_test_not_in_train)) = 0;
wolffd@0:         
wolffd@0:         subplot(2,1,2)
wolffd@0:         plot(n_train_data, lost_test_not_in_train);
wolffd@0:         
wolffd@0:         xlabel ('# training constraints');
wolffd@0:         ylabel ('% unknown constraints violated');
wolffd@0:     end
wolffd@0: end
wolffd@0: 
wolffd@0:    
wolffd@0: % --- 
wolffd@0: % write max. training success
wolffd@0: % ---
wolffd@0: mean_ok_test = [out.mean_ok_test];
wolffd@0: [val, idx] = max(mean_ok_test(1,:));
wolffd@0: fprintf(' --- Maximal training success: nr. %d, %3.2f percent. --- \n', idx, val * 100)
wolffd@0: 
wolffd@0: end