def test_measures(self):
sheet_num_universe = EvaluationSheet(self.scores, relevant=self.rel,
universe=self.num_universe)
sheet_universe = EvaluationSheet(self.scores, relevant=self.rel,
universe=self.universe)
sheet_no_universe = EvaluationSheet(self.scores, relevant=self.rel)
# Measures that don't require universe
for sheet in (sheet_num_universe, sheet_universe, sheet_no_universe):
assert_array_equal(sheet.precision(),
np.array([0, 0.5, 2 / 3, 0.5, 3 / 5,
2 / 3, 4 / 7]))
assert_array_equal(sheet.recall(),
np.array([0, 0.25, 0.5, 0.5, 0.75, 1, 1]))
# Measures that do require universe
for sheet in (sheet_num_universe, sheet_universe):
# XXX The following ones look wrong?!
expected = np.array([1 / 16, 1 / 16, 1 / 16, 1 / 8, 1 / 8, 1 / 8,
3 / 16])
assert_array_equal(sheet.fallout(), expected)
expected = np.array([4 / 19, 3 / 18, 2 / 17, 2 / 16, 1 / 15, 0, 0])
assert_array_equal(sheet.miss(), expected)
expected = np.array([0.75, 0.8, 17 / 20, 0.8, 17 / 20, 0.9,
17 / 20])
assert_array_equal(sheet.accuracy(), expected)
assert_array_equal(sheet.generality(), 0.2)
assert_raises(UndefinedError, sheet_no_universe.fallout)
assert_raises(UndefinedError, sheet_no_universe.miss)
assert_raises(UndefinedError, sheet_no_universe.accuracy)
assert_raises(UndefinedError, sheet_no_universe.generality)
def test_f_score(self):
sheet = EvaluationSheet(self.scores, relevant=self.rel)
expected = np.array([0, 2 / 6, 4 / 7, 4 / 8, 6 / 9, 8 / 10, 8 / 11])
assert_array_equal(sheet.f_score(), expected)
# $F_\beta = \frac{\beta^2 + 1 |rel \cap ret|}{\beta^2 |rel| + |ret|}$
expected = np.array([
0,
1.25 * 1 / (0.25 * 4 + 2),
1.25 * 2 / (0.25 * 4 + 3),
1.25 * 2 / (0.25 * 4 + 4),
1.25 * 3 / (0.25 * 4 + 5),
1.25 * 4 / (0.25 * 4 + 6),
1.25 * 4 / (0.25 * 4 + 7)
])
assert_array_equal(sheet.f_score(0.5), expected)
expected = np.array([
0,
5 * 1 / (4 * 4 + 2),
5 * 2 / (4 * 4 + 3),
5 * 2 / (4 * 4 + 4),
5 * 3 / (4 * 4 + 5),
5 * 4 / (4 * 4 + 6),
5 * 4 / (4 * 4 + 7)
])
assert_array_equal(sheet.f_score(2), expected)
def test_init(self):
sheet = EvaluationSheet(self.scores, relevant=self.rel)
expected = np.array([
[0, 1, 2, 2, 3, 4, 4],
[1, 1, 1, 2, 2, 2, 3],
[4, 3, 2, 2, 1, 0, 0],
[-1, -1, -1, -1, -1, -1, -1],
]).T
assert_array_equal(sheet.data, expected)
sheet = EvaluationSheet(self.scores,
relevant=self.rel,
universe=self.universe)
expected = np.array([
[0, 1, 2, 2, 3, 4, 4],
[1, 1, 1, 2, 2, 2, 3],
[4, 3, 2, 2, 1, 0, 0],
[15, 15, 15, 14, 14, 14, 13],
]).T
assert_array_equal(sheet.data, expected)
sheet = EvaluationSheet(self.scores,
relevant=self.rel,
universe=self.num_universe)
# Same expected applies as above
assert_array_equal(sheet.data, expected)
data = np.array([[1, 0, 0, 1], [1, 1, 0, 0]])
sheet = EvaluationSheet(data)
assert_array_equal(sheet.data, data)
def test_to_file_from_file(self):
data = np.array([[1, 0, 0, 1], [1, 1, 0, 0]])
sheet = EvaluationSheet(data)
with temp_file() as fname:
sheet.to_file(fname)
newsheet = EvaluationSheet.from_file(fname)
assert_array_equal(sheet.data, newsheet.data)
def test_CacheEvaluationListener():
l = CacheEvaluationListener()
scores = BaseScoresheet({1: 10, 2: 5})
ev = EvaluationSheet(scores, {1})
smokesignal.emit("evaluation_finished", ev, "d", "p")
ev2 = EvaluationSheet.from_file(l.fname)
assert_array_equal(ev.data, ev2.data)
smokesignal.clear_all()
os.unlink(l.fname)
def test_measures_with_empty_rel_and_ret(self):
sheet1 = EvaluationSheet(Scoresheet(), [], [])
sheet2 = EvaluationSheet(Scoresheet(), [], 10)
sheet3 = EvaluationSheet(Scoresheet(), [])
for sheet in (sheet1, sheet2, sheet3):
assert_raises(UndefinedError, sheet.precision)
assert_raises(UndefinedError, sheet.recall)
assert_raises(UndefinedError, sheet.f_score)
assert_raises(UndefinedError, sheet.fallout)
assert_raises(UndefinedError, sheet.miss)
assert_raises(UndefinedError, sheet.accuracy)
assert_raises(UndefinedError, sheet.generality)
def test_measures_with_empty_rel_and_ret(self):
sheet1 = EvaluationSheet(Scoresheet(), [], [])
sheet2 = EvaluationSheet(Scoresheet(), [], 10)
sheet3 = EvaluationSheet(Scoresheet(), [])
for sheet in (sheet1, sheet2, sheet3):
for method in [
"precision",
"recall",
"f_score",
"fallout",
"miss",
"accuracy",
"generality",
]:
with pytest.raises(UndefinedError):
getattr(sheet, method)()
def test_CacheEvaluationListener():
l = CacheEvaluationListener()
scores = BaseScoresheet({1: 10, 2: 5})
ev = EvaluationSheet(scores, {1})
smokesignal.emit('evaluation_finished', ev, 'd', 'p')
ev2 = EvaluationSheet.from_file(l.fname)
assert_array_equal(ev.data, ev2.data)
smokesignal.clear_all()
os.unlink(l.fname)