##### confusion matrix #####
confmat = confusion_matrix(y_true=test, y_pred=pre)
print(confmat)
##### save csv #####
pre_dict = {'Recording': [], 'Result': []}
test_dict = {'Recording': [], 'First_label': []}
count = 0
for i in range(len(pre)):
pre_dict['Recording'].append(count)
pre_dict['Result'].append(pre[i])
test_dict['Recording'].append(count)
test_dict['First_label'].append(test[i])
count += 1
pre = pd.DataFrame(pre_dict)
test = pd.DataFrame(test_dict)
test['Second_label'] = ''
test['Third_label'] = ''
pre.to_csv('./Result/1.csv', index=False)
test.to_csv('./Result/2.csv', index=False)
score_py3.score('./Result/1.csv', './Result/2.csv')
##### save process #####
log = pd.DataFrame([Accuracy, F1], index=['Accuracy', 'F1'])
log.to_csv('./Result/log.csv')
##### save figure #####
if Args.show_plot:
plt.ioff()
plt.savefig("./Result/result.jpg")
plt.show()
print('Epoch: ', epoch, '| train loss: %.4f' % loss.data,
'| train accuray: %.2f' % accuracy_train,
'| test accuracy: %.2f' % accuracy_test)
print('End Training')
pre = pred_test.data.cpu().numpy() + 1
# pre = np.array(pred_test) + 1
test = test_y.data.cpu().numpy() + 1
# test = np.array(test_y.data) + 1
pre_dict = {'Recording': [], 'Result': []}
test_dict = {'Recording': [], 'First_label': []}
count = 0
for i in range(len(pre)):
pre_dict['Recording'].append(count)
pre_dict['Result'].append(pre[i])
test_dict['Recording'].append(count)
test_dict['First_label'].append(test[i])
count += 1
pre = pd.DataFrame(pre_dict)
test = pd.DataFrame(test_dict)
# %%
test['Second_label'] = ''
test['Third_label'] = ''
# %%
pre.to_csv('1.csv', index=False)
test.to_csv('2.csv', index=False)
score('1.csv', '2.csv')
model = load_model("best_model.109-0.87.h5")
pre_lists = pd.read_csv(path_test_ref)
print(pre_lists.head())
pre_lists = np.array(pre_lists)
pre_datas = np.array([get_feature(item, path_test) for item in pre_lists[:,0]])
#for Sequential model
#pre_result = model.predict_classes(pre_datas)
#for other model
pre_r = model.predict(pre_datas)
pre_result = np.argmax(model.predict(pre_datas), axis=1)
print(pre_result.shape)
result_label = [x+1 for x in pre_result]
df1 = np.array([pre_lists[:,0]]).T
df2 = np.array([result_label]).T
df = np.hstack((df1, df2))
header = np.array(['Recording', 'Result'])
answer = np.vstack((header,df))
dataframe = pd.DataFrame(answer)
dataframe.to_csv(path_test_answer, header=False)
print("predict finish")
#Performance Evaluation
score(path_test_answer, path_test_ref)