HyperLogLog is an algorithm for estimating the cardinality of a set. The HyperLogLog strategy has several nice properties:
- It is near-optimal in its estimation ability
- allows you some coarse tuning on the amount of standard error you can tolerate
- The data structures that are used for the estimation are fast, easily compressed and stored, and can be recombined to provide estimates of both the union and intersection of multiple sets
The API is broken out into 2 pieces, the HyperBuilder and the HyperEstimator. This is done for clarity as well as performance optimizations in the future.
gem install hyperloglog
require 'hyperloglog'
# Build a new estimator
builder = HyperBuilder.new
0.upto(100).each{|user_id| builder.offer(user_id.to_s)}
# Read an estimator from bytes on disk
estimator = HyperEstimator.new(File.read('bytes.txt'))
# Estimate the union of our two sources
estimate = HyperEstimator.estimate(builder.estimator, estimator)
# puts estimate
# => 147
Murmur3 https://github.com/PeterScott/murmur3
EWAHBoolArray https://github.com/lemire/EWAHBoolArray