def test_kmeans(self):
with open('s2/tests/datasets/iris.pkl', mode='rb') as f:
expected_predictions = pickle.load(f)
kmeans = KMeans(K=3, metric='euclidean', vis_dims=0, seed=self.seed, name='test')
predictions = kmeans.fit_predict(self.dataset)
self.assertEqual(predictions, expected_predictions)
for nb_points in range(5000,46000,2500):
x.append(nb_points)
print(nb_points)
X = dataset.sample(n=nb_points)
y = [0] * nb_points
fit_times['K-Means'][nb_points] = []
fit_times['K-Means++'][nb_points] = []
fit_times['GK-Means'][nb_points] = []
fit_times['IF K-Means'][nb_points] = []
for count in range(0, 10):
print('\t{}'.format(count))
print('\tK-Means Random Initialization')
start_time = time.clock()
kmeans = KMeans(15, X=X.values, Y=y, name='NORM-10')
kmeans.find_centers()
end_time = time.clock()
fit_times['K-Means'][nb_points].append(end_time-start_time)
print('\tK-Means++')
start_time = time.clock()
kpp = KPlusPlus(15, X=X.values, Y=y, name='NORM-10')
kpp.init_centers()
kpp.find_centers(method='++')
end_time = time.clock()
fit_times['K-Means++'][nb_points].append(end_time-start_time)
print('\tK-Means Graph')
start_time = time.clock()
kmeansgraph = KMeansGraph(15, X=X, Y=y, name='NORM-10')
return temp
if __name__ == '__main__':
HOST = sys.argv[1]
PORT = int(sys.argv[2])
BROKER_HOST = sys.argv[3]
BROKER_PORT = int(sys.argv[4])
print("Initializing the web server & MQTT broker")
print("BROKER HOST {}".format(BROKER_HOST))
print("BROKER PORT {}".format(BROKER_PORT))
print("HOST {}".format(HOST))
print("PORT {}".format(PORT))
# Initialization of data postprocessing and ML algorithm
kmeans = KMeans(k=3)
kmeans.load('algorithms/model/{}.pickle'.format(MODEL_NAME))
# Calibrate the sensors
sensor = PostProcessing()
sensor.load_gyro_calibration(
file_path="algorithms/calibration/calibration_values.txt")
### Establishing code for the broker
# Establish the broker service
def on_connect(client, userdata, flags, rc):
print("Connected with result code " + str(rc))
client.subscribe("esys/HeadAid/sensor")
def on_message(client, userdata, msg):
# Get the raw data
def test_predict_before_fit_throws_error(self):
kmeans = KMeans(K=3, metric='euclidean', vis_dims=0, seed=self.seed, name='test')
with self.assertRaises(Exception):
kmeans.predict(self.dataset)
def test_kmeans_with_invalid_metric(self):
with self.assertRaises(ValueError):
_ = KMeans(K=3, metric='mahalanobis', name='test')
def test_kmeans_with_invalid_visualization_dimensions(self):
with self.assertRaises(ValueError):
_ = KMeans(K=0, name='test')
def test_kmeans_with_invalid_clusters(self):
with self.assertRaises(ValueError):
_ = KMeans(K=0, name='test')
print(train.shape)
for i in range(0, iterations):
print('Iteration {}'.format(i + 1))
# Train
train = train.rename(columns={0: 'label'})
y_train = train.label
y_train = y_train.apply(str)
X_train = train.drop("label", axis=1)
### Random initialization
print('\tK-Means Random Initialization')
start_time = time.clock()
kmeans = KMeans(args.k,
X=X_train.values,
Y=y_train.values,
name=dataset_name)
kmeans.find_centers()
end_time = time.clock()
#accuracy['kmeans'].append(kmeans.get_error_count(X_test.values, y_test.values))
method['K-Means']['phi'].append(kmeans.get_sum_distances())
method['K-Means']['fit_time'].append(end_time - start_time)
#kmeans.plot_board()
### K-means++ initialization
print('\tK-Means++')
start_time = time.clock()
kpp = KPlusPlus(args.k,
X=X_train.values,
Y=y_train.values,