def test_add_ids_random_state(self):
sketch1 = VectorOfCounts(num_buckets=8, random_seed=0)
sketch1.add_ids(range(4))
sketch2 = VectorOfCounts(num_buckets=8, random_seed=0)
sketch2.add_ids(range(4))
np.testing.assert_array_equal(
sketch1.stats,
sketch2.stats,
err_msg='Two VoC are not the same with the same random seed.')
def test_estimate_cardinality_no_clip(self):
sketch_list = []
for _ in range(3):
sketch = VectorOfCounts(num_buckets=2, random_seed=3)
sketch.add_ids([1])
sketch_list.append(sketch)
estimator = SequentialEstimator()
result = estimator(sketch_list)[0]
actual = 1.75
self.assertEqual(result, actual)
def test_merge_no_clip(self):
sketch_list = []
for _ in range(2):
sketch = VectorOfCounts(num_buckets=2, random_seed=2)
sketch.add_ids([1])
sketch_list.append(sketch)
pairwise_estimator = PairwiseEstimator()
merged = pairwise_estimator.merge(sketch_list[0], sketch_list[1])
np.testing.assert_array_equal(np.sort(merged.stats), np.array([0,
1.5]))
def test_has_full_intersection(self):
pairwise_estimator = PairwiseEstimator()
this = VectorOfCounts(num_buckets=64, random_seed=2)
this.add_ids(range(100))
that = VectorOfCounts(num_buckets=64, random_seed=2)
that.add_ids(range(100))
intersection_cardinality = pairwise_estimator._intersection(this, that)
self.assertTrue(
pairwise_estimator.has_full_intersection(intersection_cardinality,
this, that))
def test_has_zero_intersection(self):
pairwise_estimator = PairwiseEstimator()
this = VectorOfCounts(num_buckets=64, random_seed=2)
this.add_ids(range(100))
# Clip relies on hypothesis testing and hence requires a minimum size
that = VectorOfCounts(num_buckets=64, random_seed=2)
that.add_ids(range(100, 200))
intersection_cardinality = pairwise_estimator._intersection(this, that)
self.assertTrue(
pairwise_estimator.has_zero_intersection(intersection_cardinality,
this, that))
def test_estimate_cardinality_with_clip(self):
base_sketch = VectorOfCounts(num_buckets=64, random_seed=3)
base_sketch.add_ids(range(100))
sketch_list_a = [base_sketch]
sketch_list_b = [base_sketch]
for _ in range(3):
empty_sketch = VectorOfCounts(num_buckets=64, random_seed=3)
sketch_list_a.append(empty_sketch) # add empty sketch
sketch_list_b.append(base_sketch) # add same sketch
estimator = SequentialEstimator(clip=True)
result_a = estimator(sketch_list_a)
result_b = estimator(sketch_list_b)
self.assertEqual(result_a, base_sketch.cardinality(),
msg='Fail to detect the no-intersection case.')
self.assertEqual(result_b, base_sketch.cardinality(),
msg='Fail to detect the full-intersection case.')
def test_merge_with_clip(self):
this_sketch = VectorOfCounts(num_buckets=64, random_seed=2)
this_sketch.add_ids(range(100))
# First test no intersection
that_sketch = VectorOfCounts(num_buckets=64, random_seed=2)
that_sketch.add_ids(range(100, 200))
pairwise_estimator = PairwiseEstimator(clip=True)
merged = pairwise_estimator.merge(this_sketch, that_sketch)
np.testing.assert_array_equal(
x=merged.stats, y=this_sketch.stats + that_sketch.stats,
err_msg='Fail to detect the no-intersection case.')
# Then test full intersection
that_sketch = VectorOfCounts(num_buckets=64, random_seed=2)
that_sketch.add_ids(range(100))
merged = pairwise_estimator.merge(this_sketch, that_sketch)
np.testing.assert_array_equal(
x=merged.stats, y=this_sketch.stats,
err_msg='Fail to detect the full-intersection case.')
def test_add_ids_multiple_times(self):
sketch = VectorOfCounts(num_buckets=8, random_seed=0)
sketch.add_ids([1])
with self.assertRaises(AssertionError):
sketch.add_ids([1])
def test_add_ids(self):
sketch = VectorOfCounts(num_buckets=8, random_seed=0)
sketch.add_ids([1])
self.assertEqual(sketch.cardinality(), 1)