HyperLogLog implementation for Ruby. Originally written in pure Ruby, many pieces were rewritten in C for increased performance.
HyperLogLog stores an estimation of the cardinality of a set. It can be merged with other HyperLogLog instances.
hll = Hyperll::HyperLogLog.new(10)
hll.offer(1)
hll.offer(2)
hll.offer(3)
hll.cardinality # => 3
hll2 = Hyperll::HyperLogLog.new(10)
hll2.offer(3)
hll2.offer(4)
hll2.offer(5)
hll.cardinality # => 3
merged = Hyperll::HyperLogLog.new(10)
merged.merge(hll, hll2)
merged.cardinality # => 5HyperLogLog can be serialized to a binary string. It is compatible with the binary format from the Java stream-lib library.
hll = Hyperll::HyperLogLog.new(4)
hll.offer(1)
hll.offer(2)
hll.offer(3)
hll.serialize # => "\x00\x00\x00\x04\x00\x00\x00\f\x02\x00\x00\x00\x00\x00\x88\x00\x00\x00\x00\x00"
hll2 = Hyperll::HyperLogLog.unserialize("\x00\x00\x00\x04\x00\x00\x00\f\x02\x00\x00\x00\x00\x00\x88\x00\x00\x00\x00\x00")
hll2.cardinality # => 3