Crate sketches 

Probabilistic data structures for scalable approximate analytics.

The crate currently exposes:

• minmax_sketch::MinMaxSketch for approximate frequency estimation.
• hyperloglog::HyperLogLog for approximate cardinality estimation.
• jacard for approximate set overlap/Jaccard helpers on HyperLogLog.
• bloom_filter::BloomFilter for approximate set membership checks.
• count_sketch::CountSketch for signed approximate frequency estimation.
• space_saving::SpaceSaving for approximate heavy hitters.
• kll::KllSketch for approximate quantiles.
• tdigest::TDigest for tail-friendly quantiles.
• cuckoo_filter::CuckooFilter for membership with deletions.
• minhash::MinHash for approximate Jaccard estimation.
• lsh_minhash::MinHashLshIndex for approximate nearest-neighbor lookup.
• reservoir_sampling::ReservoirSampling for uniform stream sampling.

Modules

bloom_filter

Bloom filter for approximate set membership.

count_sketch

Count Sketch for approximate frequency estimation.

cuckoo_filter

Cuckoo filter for approximate set membership with deletions.

hyperloglog

HyperLogLog cardinality estimator.

jacard

Jaccard similarity trait shared by sketch implementations.

kll

KLL sketch for approximate quantile queries.

lsh_minhash

MinHash banding LSH index for approximate nearest-neighbor candidate search.

minhash

MinHash sketch for approximate Jaccard similarity.

minmax_sketch

MinMax sketch for approximate frequency counting.

reservoir_sampling

Reservoir sampling for uniform samples from streaming data.

space_saving

Space-Saving sketch for approximate heavy hitters.

tdigest

t-digest for approximate quantiles, especially in distribution tails.

Enums

SketchError

Errors returned by sketch constructors and merge operations.