def __compute__(fnc, classes, arguments, classification, version):
if classification is None:
if version is Average.micro:
# Micro-average
numerator, denominator = 0, 0
for classification in classes:
arguments['classification'] = classification
arguments['return_parts'] = True
class_numerator, class_denominator = fnc(**arguments)
numerator += class_numerator
denominator += class_denominator
if numerator == 0:
return 0
else:
return Decimal(numerator) / denominator
elif version is Average.macro:
# Macro-average
values = list()
for classification in classes:
arguments['classification'] = classification
value = fnc(**arguments)
values.append(value)
return mean(values)
else:
raise SegmentationMetricError('Unrecognized type of averaging;\
expected Average.micro or Average.macro')
else:
return fnc(**arguments)
def __compute__(fnc, classes, arguments, classification, version):
if classification is None:
if version is Average.micro:
# Micro-average
numerator, denominator = 0, 0
for classification in classes:
arguments['classification'] = classification
arguments['return_parts'] = True
class_numerator, class_denominator = fnc(**arguments)
numerator += class_numerator
denominator += class_denominator
if numerator == 0:
return 0
else:
return Decimal(numerator) / denominator
elif version is Average.macro:
# Macro-average
values = list()
for classification in classes:
arguments['classification'] = classification
value = fnc(**arguments)
values.append(value)
return mean(values)
else:
raise SegmentationMetricError('Unrecognized type of averaging;\
expected Average.micro or Average.macro')
else:
return fnc(**arguments)
def summarize(pairs):
'''
Takes a list of values and returns the mean, standard deviation, variance, standard error, and number of values.
:param pairs: List of numerical values
:type pairs: list
'''
return mean(pairs.values()), std(pairs.values()), var(pairs.values()), \
stderr(pairs.values()), len(pairs)
def summarize(pairs):
'''
Takes a list of values and returns the mean, standard deviation, variance, standard error, and number of values.
:param pairs: List of numerical values
:type pairs: list
'''
return mean(pairs.values()), std(pairs.values()), var(pairs.values()), \
stderr(pairs.values()), len(pairs)
def __compute_window_size__(reference, fnc_round, boundary_format):
'''
Compute a window size from a dict of segment masses.
:param masses: A dict of segment masses.
:type masses: dict
'''
all_masses = list()
# Define fnc
def __list_coder_masses__(inner_coder_masses):
'''
Recursively collect all masses.
:param inner_coder_masses: Either a dict of dicts, or dict of a list of
masses.
:type inner_coder_masses: dict or list
'''
if hasattr(inner_coder_masses, 'items'):
for cur_inner_coder_masses in inner_coder_masses.values():
__list_coder_masses__(cur_inner_coder_masses)
elif hasattr(inner_coder_masses, '__iter__') and not isinstance(inner_coder_masses, str):
all_masses.extend(inner_coder_masses)
else:
raise SegmentationMetricError('Expected either a dict-like \
collection of segmentations or a segmentation as a list-like object')
if boundary_format == BoundaryFormat.position:
reference = convert_positions_to_masses(reference)
# Recurse and list all masses
__list_coder_masses__(reference)
# Convert to floats
all_masses = [Decimal(mass) for mass in all_masses]
# Calculate
avg = mean(all_masses) / Decimal('2')
window_size = int(fnc_round(avg))
return window_size if window_size > 1 else 2
def test_mean_0(self):
'''
Tests population mean.
'''
self.assertEqual(0, mean([]))
def test_mean(self):
'''
Tests population mean.
'''
self.assertEqual(5, mean([2, 4, 4, 4, 5, 5, 7, 9]))
def test_mean_0(self):
'''
Tests population mean.
'''
self.assertEqual(0, mean([]))
def test_mean(self):
'''
Tests population mean.
'''
self.assertEqual(5, mean([2, 4, 4, 4, 5, 5, 7, 9]))
