def performance_dump(PARAMS,
PtdLabels,
GroundTruths,
labels,
info='',
fName_suffix=''):
ConfMat, precision, recall, fscore = misc.getPerformance(
PtdLabels, GroundTruths, labels)
accuracy = np.round(np.sum(np.diag(ConfMat)) / np.sum(ConfMat), 4)
print('Total data performance: ', fscore)
print(ConfMat)
if len(labels) == 2:
classnames = ['neg', 'pos']
else:
classnames = ['mu', 'sp', 'spmu']
res_dict = {}
res_dict['0'] = 'feature_name:' + PARAMS['featName'][PARAMS['Model']]
res_dict['1'] = 'model:' + PARAMS['Model']
ln = 2
if not info == '':
res_dict[str(ln)] = info
ln += 1
res_dict[str(ln)] = 'loss:--'
ln += 1
res_dict[str(ln)] = 'accuracy:' + str(accuracy)
ln += 1
res_dict[str(ln)] = 'Prec_' + classnames[0] + ':' + str(precision[0])
ln += 1
res_dict[str(ln)] = 'Rec_' + classnames[0] + ':' + str(recall[0])
ln += 1
res_dict[str(ln)] = 'F1_' + classnames[0] + ':' + str(fscore[0])
ln += 1
res_dict[str(ln)] = 'Prec_' + classnames[1] + ':' + str(precision[1])
ln += 1
res_dict[str(ln)] = 'Rec_' + classnames[1] + ':' + str(recall[1])
ln += 1
res_dict[str(ln)] = 'F1_' + classnames[1] + ':' + str(fscore[1])
if len(labels) == 3:
ln += 1
res_dict[str(ln)] = 'Prec_' + classnames[2] + ':' + str(precision[2])
ln += 1
res_dict[str(ln)] = 'Rec_' + classnames[2] + ':' + str(recall[2])
ln += 1
res_dict[str(ln)] = 'F1_' + classnames[2] + ':' + str(fscore[2])
ln += 1
res_dict[str(ln)] = 'F1_avg:' + str(np.round(np.mean(fscore), 4))
misc.print_results(PARAMS, fName_suffix, res_dict)
feature_indexes_dict_keys[classifier_num] + '_iter' +
str(numIter))
kwargs = {
'0': 'epochs:' + str(Train_Params['epochs']),
'1': 'batch_size:' + str(Train_Params['batch_size']),
'2': 'learning_rate:' + str(Train_Params['learning_rate']),
'3':
'training_time:' + str(Train_Params['trainingTimeTaken']),
'4': 'loss:' + str(Test_Params['loss']),
'5': 'performance:' + str(Test_Params['performance']),
'6': 'F_score_mu:' + str(Test_Params['fscore'][0]),
'7': 'F_score_sp:' + str(Test_Params['fscore'][1]),
'8': 'F_score_avg:' + str(Test_Params['fscore'][2]),
}
misc.print_results(PARAMS, **kwargs)
print('Performance: ', Test_Params['performance'],
Test_Params['ConfMat'])
print('Avg. F1-score: ', Test_Params['fscore'][-1])
print('Training time taken: ',
Train_Params['trainingTimeTaken'],
Test_Params['testingTimeTaken'])
elif PARAMS['clFunc'] == 'SVM-Ensemble':
''' ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ '''
PARAMS['modelName'] = PARAMS[
'opDir'] + '/SVM_' + feature_indexes_dict_keys[
classifier_num] + '_iter' + str(
PARAMS['iter']) + '.pkl'
Train_Params, Test_Params = SVM.grid_search_svm(
PARAMS, data_dict_Classifier_Part)
i = 0
if (PARAMS['clFunc']=='DNN-Ensemble') or (PARAMS['clFunc']=='CNN-Ensemble'):
for key in Ensemble_Test_Params['individual_performances'].keys():
kwargs[str(i+1)] = key + ':' + str(Ensemble_Test_Params['individual_performances'][key]['fscore'][2])
i += 1
elif (PARAMS['clFunc']=='SVM-Ensemble') or (PARAMS['clFunc']=='NB-Ensemble'):
for key in All_Test_Params.keys():
kwargs[str(i+1)] = key + ':' + str(All_Test_Params[key]['fscore'][2])
i += 1
kwargs['8'] = 'Accuracy:'+str(Ensemble_Test_Params['accuracy_Ensemble'])
kwargs['9'] = 'F_score_mu:'+str(Ensemble_Test_Params['fscore_Ensemble'][0])
kwargs['10'] = 'F_score_sp:'+str(Ensemble_Test_Params['fscore_Ensemble'][1])
kwargs['11'] = 'F_score_avg:'+str(Ensemble_Test_Params['fscore_Ensemble'][2])
misc.print_results(PARAMS, '', **kwargs)
else:
if PARAMS['clFunc']=='DNN-Ensemble':
Ensemble_Test_Params = DNN.test_dnn_ensemble_noise(PARAMS, Ensemble_Train_Params)
elif PARAMS['clFunc']=='CNN-Ensemble':
Ensemble_Test_Params = CNN.test_cnn_ensemble_noise(PARAMS, Ensemble_Train_Params)
kwargs = {}
kwargs['0'] = 'feature_type:'+feature_type
i = 1
for dB in PARAMS['noise_dB_range']:
kwargs[str(i)] = str(dB)+'dB_ACC:'+str(Ensemble_Test_Params['accuracy_Ensemble'][dB])
kwargs[str(i+1)] = str(dB)+'dB_AVG_F1:'+str(Ensemble_Test_Params['fscore_Ensemble'][dB][2])
i += 2
misc.print_results(PARAMS, 'noise_exp', **kwargs)
print('Test_Params: ', Test_Params.keys())
print(Test_Params['precision_annot'], Test_Params['recall_annot'],
Test_Params['fscore_annot'])
res_dict = {}
res_dict['0'] = 'SMR:Annot'
res_dict['1'] = Test_Params['metric_names'][0] + ':' + str(
Test_Params['metrics'][0])
res_dict['2'] = Test_Params['metric_names'][1] + ':' + str(
Test_Params['metrics'][1])
res_dict['3'] = 'Prec_mu:' + str(Test_Params['precision_annot'][0])
res_dict['4'] = 'Rec_mu:' + str(Test_Params['recall_annot'][0])
res_dict['5'] = 'F1_mu:' + str(Test_Params['fscore_annot'][0])
res_dict['6'] = 'Prec_sp:' + str(Test_Params['precision_annot'][1])
res_dict['7'] = 'Rec_sp:' + str(Test_Params['recall_annot'][1])
res_dict['8'] = 'F1_sp:' + str(Test_Params['fscore_annot'][1])
if len(PARAMS['classes']) == 3:
res_dict['9'] = 'Prec_spmu:' + str(
Test_Params['precision_annot'][2])
res_dict['10'] = 'Rec_spmu:' + str(
Test_Params['recall_annot'][2])
res_dict['11'] = 'F1_spmu:' + str(
Test_Params['fscore_annot'][2])
misc.print_results(PARAMS, '', res_dict)
Train_Params = None
Test_Params = None
if PARAMS['use_GPU']:
reset_TF_session()