def walking_speed(self, labels, timestamps, timespan, overlap, fs):
"""
Wrapper to output the average walking speed.
Parameters
----------
labels
Vector of labels.
timestamps
Vector of timestamps.
timespan
Duration of metric calculation. Either 86400 (daily) or 3600 (hourly)
overlap
Amount of overlap for the calculation of the transfers.
fs
Sampling frequency.
Returns
-------
metr.number_of_label_changes_per_window(labels, timestamps)
Output the average walking speed.
"""
metr = Metrics(timestamps, timespan, overlap, fs)
container = metr.speed(labels, timestamps, self.adjecency)
# TODO Why do we want the absolute values for speed, and these seem not
# to be applied to the specified window, instead it is the raw values
# after the apply metrics is called
return np.abs(container), np.nanmean(np.abs(container)), np.nanmax(
np.abs(container))
def number_of_unique_locations(self, labels, timestamps, timespan, overlap,
fs):
"""
Wrapper to output the number of unique locations per window.
Parameters
----------
labels
Vector of labels.
timestamps
Vector of timestamps.
timespan
Duration of metric calculation. Either 86400 (daily) or 3600 (hourly)
overlap
Amount of overlap for the calculation of the transfers.
fs
Sampling frequency.
Returns
-------
metr.number_of_label_changes_per_window(labels, timestamps)
Output the number of unique locations per window.
"""
metr = Metrics(timestamps, timespan, overlap, fs)
container = metr.duration_of_labels_per_window(labels, timestamps)
container = self.label_mappings(container, is_duration=False)
return container
def duration_in_location(self, labels, timestamps, timespan, overlap, fs,
location):
"""
Wrapper to output the duration studying.
Parameters
----------
labels
Vector of labels.
timestamps
Vector of timestamps.
timespan
Duration of metric calculation. Either 86400 (daily) or 3600 (hourly)
overlap
Amount of overlap for the calculation of the transfers.
fs
Sampling frequency.
location
Location in which the duration is computed
Returns
-------
metr.number_of_label_changes_per_window(labels, timestamps)
Output the duration studying.
"""
metr = Metrics(timestamps, timespan, overlap, fs)
container = metr.duration_of_labels_per_window(labels, timestamps)
container = self.label_mappings(container,
is_duration=False,
label_to_extract=location)
return container
def duration_activity(self, labels, timestamps, timespan, overlap, fs,
activity):
"""
Wrapper to output the duration walking.
Parameters
----------
labels
Vector of labels.
timestamps
Vector of timestamps.
timespan
Duration of metric calculation. Either 86400 (daily) or 3600 (hourly)
overlap
Amount of overlap for the calculation of the transfers.
fs
Sampling frequency.
Returns
-------
metr.number_of_label_changes_per_window(labels, timestamps)
Output the duration walking.
"""
metr = Metrics(timestamps, timespan, overlap, fs)
container = metr.duration_of_labels_per_window(labels, timestamps)
container = self.label_mappings(container, True, activity)
return container
def room_transfers(self, labels, timestamps, timespan, overlap, fs):
"""
Wrapper to output the room transfers.
Parameters
----------
labels
Vector of labels.
timestamps
Vector of timestamps.
timespan
Duration of metric calculation. Either 86400 (daily) or 3600 (hourly)
overlap
Amount of overlap for the calculation of the transfers.
fs
Sampling frequency.
Returns
-------
metr.number_of_label_changes_per_window(labels, timestamps)
Output the room transfers.
"""
metr = Metrics(timestamps, timespan, overlap, fs)
container = metr.number_of_label_changes_per_window(labels, timestamps)
return container
def test_average_activities_per_window(self):
"""
Test average activities per window.
"""
print('Test average activities per window.')
winlength = 3
# two types of label
labels = [1, 2, 1, 1, 1, 2, 2, 2, 1]
#monotonically increasing
timestamp = [1, 2, 3, 4, 5, 6, 7, 8, 9]
groundtruth = [[[1, 0.67], [2, 0.33]], [[1, 0.67], [2,
0.33]], [[1, 1.]],
[[1, 0.67], [2, 0.33]], [[1, 0.33], [2, 0.67]],
[[2, 1.]], [[1, 0.33], [2, 0.67]]]
indicies = np.arange(len(labels))
metr = Metrics(timestamp, aggregation_duration=3, window_overlap=1)
metr.current_position = winlength
final_result = []
while True:
windowed_index = metr.slide(np.array(indicies))
windowed_labels = []
windowed_time = []
for win_idx in windowed_index:
windowed_labels = np.append(windowed_labels, labels[win_idx])
windowed_time = np.append(windowed_time, timestamp[win_idx])
if len(windowed_index) > 0 and len(windowed_index) == winlength:
result = metr.average_labels_per_window(
windowed_labels, windowed_time)
result = np.round(result, 2)
final_result.append(result)
else:
break
for x, y in zip(final_result, groundtruth):
np.testing.assert_almost_equal(x, y)
def test_speed(self):
"""
Test speed.
"""
print('Test speed.')
winlength = 9
# two types of label
labels = [1, 2, 1, 1, 1, 2, 2, 2, 1]
# monotonically increasing
timestamp = [1, 2, 3, 4, 5, 6, 7, 8, 9]
adjecency = [[0, 1], [1, 0]]
groundtruth = np.array([0, 1, 0, 0, 1, 0, 0, 1])
indicies = np.arange(len(labels))
metr = Metrics(timestamp, aggregation_duration=9, window_overlap=1)
metr.current_position = winlength
final_result = []
while True:
windowed_index = metr.slide(np.array(indicies))
windowed_labels = []
windowed_time = []
for win_idx in windowed_index:
windowed_labels = np.append(windowed_labels, labels[win_idx])
windowed_time = np.append(windowed_time, timestamp[win_idx])
if len(windowed_index) > 0 and len(
windowed_index) == winlength - 1:
result = metr.speed(windowed_labels, windowed_time, adjecency)
result = np.round(result, 2)
final_result.append(result)
else:
break
for x, y in zip(np.squeeze(final_result), groundtruth):
np.testing.assert_almost_equal(x, y)