def run_test(training_subjects,
test_subjects,
C=1,
kernel='rbf',
degree=1,
gamma='auto',
data_source='',
permutate_xyz=False,
activities=None,
show_confusion=False):
X_train, Y_train = data_util.load_training_data(
training_subjects,
data_source + '_' + ''.join(training_subjects) + '_svmloo4_' +
CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities,
permutate_xyz=permutate_xyz)
X_test, Y_test = data_util.load_testing_data(
test_subjects,
data_source + '_' + ''.join(training_subjects) + '_svmloo4_' +
CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities)
print('X_train size: {}'.format(len(X_train)))
print('X_test size: {}'.format(len(X_test)))
print()
accuracy_results = []
fscore_results = []
model = SVC(C=C,
kernel=kernel,
gamma=gamma,
degree=degree,
random_state=int(time.time()))
model.fit(X_train, Y_train)
predictions = model.predict(X_test)
accuracy = accuracy_score(Y_test, predictions)
accuracy_results.append(accuracy)
fscore = f1_score(Y_test, predictions, average='weighted')
fscore_results.append(fscore)
if show_confusion:
cm = confusion_matrix(Y_test, predictions)
c_matrix.append(cm)
return accuracy, fscore
def run_test(n_estimators=50,
data_source='',
activities=None,
permutate_xyz=False,
show_confusion=False):
X_train, Y_train = data_util.load_training_data(
CONFIG.TRAINING_DATA_SOURCE_SUBJECT,
data_source + '_' + ''.join(CONFIG.TRAINING_DATA_SOURCE_SUBJECT) +
'_rfloo_' + CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities,
permutate_xyz=permutate_xyz)
X_test, Y_test = data_util.load_testing_data(
CONFIG.TESTING_DATA_SOURCE_SUBJECT,
data_source + '_' + ''.join(CONFIG.TRAINING_DATA_SOURCE_SUBJECT) +
'_rfloo_' + CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities)
print('X_train size: {}'.format(len(X_train)))
print('X_test size: {}'.format(len(X_test)))
print()
accuracy_results = []
fscore_results = []
model = RandomForestClassifier(n_estimators=n_estimators, n_jobs=-1)
model.fit(X_train, Y_train)
predictions = model.predict(X_test)
accuracy = accuracy_score(Y_test, predictions)
accuracy_results.append(accuracy)
fscore = f1_score(Y_test, predictions, average='weighted')
fscore_results.append(fscore)
if show_confusion:
cm = confusion_matrix(Y_test, predictions)
plt.figure()
plot_util.plot_confusion_matrix(cm, [
activity_encoding.INT_TO_ACTIVITY_MAPPING[i] for i in sorted([
activity_encoding.ACTIVITY_TO_INT_MAPPING[a]
for a in activities
])
])
plt.show()
return accuracy, fscore
def run_test(n_neighbors=50,
data_source='',
activities=None,
permutate_xyz=False):
X_train, Y_train = data_util.load_training_data(
CONFIG.TRAINING_DATA_SOURCE_SUBJECT,
CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities,
permutate_xyz=permutate_xyz)
X_test, Y_test = data_util.load_testing_data(
CONFIG.TESTING_DATA_SOURCE_SUBJECT,
CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities)
print('X_train size: {}'.format(len(X_train)))
print('X_test size: {}'.format(len(X_test)))
print()
accuracy_results = []
fscore_results = []
model = KNeighborsClassifier(n_neighbors=n_neighbors,
n_jobs=-1,
weights='uniform')
model.fit(X_train, Y_train)
predictions = model.predict(X_test)
accuracy = accuracy_score(Y_test, predictions)
accuracy_results.append(accuracy)
fscore = f1_score(Y_test, predictions, average='weighted')
fscore_results.append(fscore)
# cm = confusion_matrix(
# Y_test,
# predictions
# )
#
# plt.figure()
# plot_util.plot_confusion_matrix(
# cm,
# [activity_encoding.INT_TO_ACTIVITY_MAPPING[i] for i in sorted([activity_encoding.ACTIVITY_TO_INT_MAPPING[a] for a in activities])]
# )
# plt.show()
return accuracy, fscore
def run_training(c=100, gamma=0.5, kernel='rbf', data_source=''):
print('Training SVM model for Real Time Monitoring..')
print('Cost: {}'.format(c))
print('Gamma: {}'.format(gamma))
print('Kernel: {}'.format(kernel))
print('Data Source: {}'.format(data_source))
print()
print()
print()
X_train, Y_train = data_util.load_training_data(
CONFIG.REAL_TIME_MONITORING_TRAINING_DATA_SOURCE_SUBJECT,
CONFIG.MODEL_NAMES['real_time_monitoring_minmax_scaler'],
source=data_source)
model = SVC(C=c, kernel=kernel, gamma=gamma)
model.fit(X_train, Y_train)
return model
def run_training(n_estimators=50, data_source=''):
print('Training RF model for Real Time Monitoring..')
print('Number of Estimators: {}'.format(n_estimators))
print('Data Source: {}'.format(data_source))
print('Training Subjects: {}'.format(CONFIG.REAL_TIME_MONITORING_TRAINING_DATA_SOURCE_SUBJECT))
print()
print()
print()
X_train, Y_train = data_util.load_training_data(
CONFIG.REAL_TIME_MONITORING_TRAINING_DATA_SOURCE_SUBJECT,
CONFIG.MODEL_NAMES['real_time_monitoring_minmax_scaler'],
source=data_source,
)
model = RandomForestClassifier(n_estimators=n_estimators, n_jobs=-1)
model.fit(X_train, Y_train)
return model
def run_test(training_subjects, test_subjects, n_estimators=50, data_source='', activities=None, permutate_xyz=False, show_confusion=False):
X_train, Y_train = data_util.load_training_data(
training_subjects,
data_source + '_' + ''.join(training_subjects) + '_rfloo2_' + CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities,
permutate_xyz=permutate_xyz
)
X_test, Y_test = data_util.load_testing_data(
test_subjects,
data_source + '_' + ''.join(training_subjects) + '_rfloo2_' + CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities
)
print('X_train size: {}'.format(len(X_train)))
print('X_test size: {}'.format(len(X_test)))
print()
accuracy_results = []
fscore_results = []
model = RandomForestClassifier(n_estimators=n_estimators, n_jobs=-1, random_state=int(time.time()))
model.fit(X_train, Y_train)
predictions = model.predict(X_test)
accuracy = accuracy_score(Y_test, predictions)
accuracy_results.append(accuracy)
fscore = f1_score(Y_test, predictions, average='weighted')
fscore_results.append(fscore)
if show_confusion:
cm = confusion_matrix(Y_test, predictions)
c_matrix.append(cm)
return accuracy, fscore
def run_test(C=10,
kernel='rbf',
degree=1,
gamma='auto',
data_source='',
permutate_xyz=False,
activities=None):
print('Data Source: {}'.format(data_source))
print('Cost: {}'.format(C))
print('Gamma: {}'.format(gamma))
print('Sampling Frequency: {}'.format(CONFIG.SAMPLING_FREQUENCY))
print('Window Size: {}'.format(CONFIG.WINDOW_SIZE))
print('Training Subjects: {}'.format(CONFIG.TRAINING_DATA_SOURCE_SUBJECT))
print('Testing Subjects: {}'.format(CONFIG.TESTING_DATA_SOURCE_SUBJECT))
X_train, Y_train = data_util.load_training_data(
CONFIG.TRAINING_DATA_SOURCE_SUBJECT,
data_source + '_' + ''.join(CONFIG.TRAINING_DATA_SOURCE_SUBJECT) +
'_svmloo_' + CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities,
permutate_xyz=permutate_xyz)
X_test, Y_test = data_util.load_testing_data(
CONFIG.TESTING_DATA_SOURCE_SUBJECT,
data_source + '_' + ''.join(CONFIG.TRAINING_DATA_SOURCE_SUBJECT) +
'_svmloo_' + CONFIG.MODEL_NAMES['minmax_scaler'],
source=data_source,
activities=activities)
print('X_train size: {}'.format(len(X_train)))
print('X_test size: {}'.format(len(X_test)))
print()
accuracy_results = []
fscore_results = []
model = SVC(C=C,
kernel=kernel,
gamma=gamma,
degree=degree,
random_state=int(time.time()))
model.fit(X_train, Y_train)
predictions = model.predict(X_test)
accuracy = accuracy_score(Y_test, predictions)
accuracy_results.append(accuracy)
fscore = f1_score(Y_test, predictions, average='weighted')
fscore_results.append(fscore)
# cm = confusion_matrix(
# Y_test,
# predictions
# )
#
# plt.figure()
# plot_util.plot_confusion_matrix(
# cm,
# [activity_encoding.INT_TO_ACTIVITY_MAPPING[i] for i in sorted([activity_encoding.ACTIVITY_TO_INT_MAPPING[a] for a in activities])]
# )
# plt.show()
return accuracy, fscore
accuracy = accuracy_score(Y_test, predictions)
fscore = f1_score(Y_test, predictions, average='weighted')
print()
print()
print('Sampling Frequency: {}'.format(CONFIG.SAMPLING_FREQUENCY))
print('Window Size: {}'.format(CONFIG.WINDOW_SIZE))
print('Accuracy: {}'.format(accuracy))
print('F1 Score: {}'.format(fscore))
print()
print('==================================================================')
if __name__ == '__main__':
X_train, Y_train = data_util.load_training_data(
CONFIG.TRAINING_DATA_SOURCE_SUBJECT,
CONFIG.MODEL_NAMES['minmax_scaler'],
onehot=True)
X_test, Y_test = data_util.load_testing_data(
CONFIG.TESTING_DATA_SOURCE_SUBJECT,
CONFIG.MODEL_NAMES['minmax_scaler'],
onehot=False)
alpha = [0.01]
max_epoch = [2000]
n_hidden_layer = [1]
n_neuron_hidden_layer = [100]
for a in alpha:
for epoch in max_epoch:
for layer in n_hidden_layer:
