def test_frequency():
cms = CountMinSketch(4, 100)
cms.add("test")
assert cms.frequency("test") == 1, "Can't find recently added element"
assert cms.frequency("test_test") == 0, "False positive detected"
21,
9,
29,
6,
5,
2,
3,
1,
16,
17,
15,
5,
3,
6,
9,
12,
]
if __name__ == '__main__':
cms = CountMinSketch(4, 100)
print(cms)
print("CMS uses {} bytes in the memory".format(cms.sizeof()))
for digit in DATASET:
cms.add(digit)
for digit in sorted(set(DATASET)):
print("Element: {}. Freq.: {}, Est. Freq.: {}".format(
digit, DATASET.count(digit), cms.frequency(digit)))
def test_add():
cms = CountMinSketch(4, 100)
for word in ["test", 1, {"hello": "world"}]:
cms.add(word)
assert cms.frequency(word) == 1, "Can't find frequency for element"