def test_get_symmetric_metric_tensor(self):
metric = krippendorffs_alpha.metrics.NominalMetric()
expected_matrix = numpy.array([[0, 1, 1, 1, 1], [1, 0, 1, 1, 1],
[1, 1, 0, 1, 1], [1, 1, 1, 0, 1],
[1, 1, 1, 1, 0]])
result_matrix = metric.get_metric_tensor([0, 1, 2, 3, 4])
testing_utils.assert_equal_tensors(expected_matrix, result_matrix)
def test_get_asymmetric_metric_tensor(self):
metric = krippendorffs_alpha.metrics.IntervalMetric()
expected_matrix = numpy.array([[0, 1, 4, 9, 16], [0, 0, 1, 4, 9],
[0, 0, 0, 1, 4], [0, 0, 0, 0, 1],
[0, 0, 0, 0, 0]])
result_matrix = metric.get_metric_tensor([0, 1, 2, 3, 4],
symmetric=False)
testing_utils.assert_equal_tensors(expected_matrix, result_matrix)
def test_get_symmetric_metric_tensor(self):
metric = krippendorffs_alpha.metrics.RatioMetric()
expected_matrix = numpy.array([[0, 1, 1, 1, 1],
[1, 0, 1 / 9, 1 / 4, 9 / 25],
[1, 1 / 9, 0, 1 / 25, 4 / 36],
[1, 1 / 4, 1 / 25, 0, 1 / 49],
[1, 9 / 25, 4 / 36, 1 / 49, 0]])
result_matrix = metric.get_metric_tensor([0, 1, 2, 3, 4])
testing_utils.assert_equal_tensors(expected_matrix, result_matrix)
def test_data_from_example_omit_unpairable(self):
expected_value_by_unit_matrix = numpy.array([
[3, 0, 0, 0, 0, 1, 0, 3, 0, 0, 2],
[0, 3, 0, 0, 4, 1, 0, 1, 4, 0, 0],
[0, 1, 4, 4, 0, 1, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 1, 4, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0],
])
actual_value_by_unit_matrix = krippendorffs_alpha._calculation._make_value_by_unit_matrix_from_data_matrix(
DATA_MATRIX, omit_unpairable=True
)
testing_utils.assert_equal_tensors(expected_value_by_unit_matrix, actual_value_by_unit_matrix)
def test_e_interval(self):
with open(os.path.join("tests", "example_E_data.tsv"), "r") as f:
input_table = csv.reader(f, delimiter="\t")
prepared_data = krippendorffs_alpha.data_converters.from_list_of_lists(
input_table=input_table,
header=True,
row_legend=True,
upper_level="observer",
value_constructor=lambda s: int(s.strip()) if s.strip() != "NULL" else None
)
testing_utils.assert_equal_tensors(DATA_MATRIX, prepared_data.answers_tensor)
actual_alpha = krippendorffs_alpha._calculation.calc_alpha(prepared_data, "interval")
assert actual_alpha == pytest.approx(0.849, abs=0.001)
def test_values_interval_ram(self):
metric = "interval"
strategy = "ram-hungry"
with open(os.path.join("tests", "example_E_data.tsv"), "r") as f:
input_table = [row for row in csv.reader(f, delimiter="\t")]
input_table = input_table[1:]
input_table = [[
int(val) if val.strip() != "NULL" else None for val in row[1:]
] for row in input_table]
users_count = len(input_table)
expected_outed_alphas = []
for user_index in range(users_count):
jackknife_table = input_table[:user_index] + input_table[
user_index + 1:]
prepared_jackknife_table = krippendorffs_alpha.data_converters.from_list_of_lists(
jackknife_table,
header=False,
row_legend=False,
upper_level="observer",
value_constructor=None,
possible_values=[1, 2, 3, 4, 5])
outer_alpha = krippendorffs_alpha._calculation.calc_alpha(
prepared_jackknife_table, metric=metric)
expected_outed_alphas.append(outer_alpha)
prepared_input_table = krippendorffs_alpha.data_converters.from_list_of_lists(
input_table,
header=False,
row_legend=False,
upper_level="observer",
value_constructor=None,
possible_values=[1, 2, 3, 4, 5])
actual_outer_alphas = krippendorffs_alpha._bootstrap.observerwise_jackknife(
prepared_data=prepared_input_table,
metric=metric,
strategy=strategy)
expected_outed_alphas = numpy.array(expected_outed_alphas)
assert len(expected_outed_alphas) == len(actual_outer_alphas), (
len(expected_outed_alphas), len(actual_outer_alphas),
actual_outer_alphas)
testing_utils.assert_equal_tensors(expected_outed_alphas,
actual_outer_alphas)
def test_wikipedia_ratio(self):
metric = "ratio"
with open(os.path.join("tests", "example_wikipedia.csv"), "r") as f:
input_table = [row for row in csv.reader(f, delimiter=",")]
input_table = input_table[1:]
input_table = [[
int(val) if val.strip() != "*" else None for val in row[1:]
] for row in input_table]
input_table = numpy.transpose(input_table).tolist()
assert isinstance(input_table, list)
expected_alphas = []
for unit_index in range(len(input_table)):
jackknife_table = input_table[:unit_index] + input_table[
unit_index + 1:]
prepared_jackknife_table = krippendorffs_alpha.data_converters.from_list_of_lists(
jackknife_table,
header=False,
row_legend=False,
upper_level="unit",
value_constructor=None,
possible_values=[1, 2, 3, 4, 5])
outer_alpha = krippendorffs_alpha._calculation.calc_alpha(
prepared_jackknife_table, metric=metric)
expected_alphas.append(outer_alpha)
prepared_input_table = krippendorffs_alpha.data_converters.from_list_of_lists(
input_table,
header=False,
row_legend=False,
upper_level="unit",
value_constructor=None,
possible_values=[1, 2, 3, 4, 5])
actual_outer_alphas = krippendorffs_alpha._bootstrap.unitwise_jackknife(
prepared_data=prepared_input_table, metric=metric)
expected_alphas = numpy.array(expected_alphas)
testing_utils.assert_equal_tensors(expected_alphas,
actual_outer_alphas)