An implementation of the HyperLogLog algorithm in Erlang
An implementation of the HyperLogLog algorithm in Erlang. Using HyperLogLog you can estimate the cardinality of very large data sets using constant memory. The relative error is 1.04 * sqrt(2^P). When creating a new HyperLogLog filter, you provide the precision P, allowing you to trade memory for accuracy. The union of two filters is lossless.