Package index • data.sketches

Quantile sketches

KLL, REQ, and t-Digest sketches for approximate quantiles, ranks, CDF, and PMF. t-Digest concentrates accuracy near the tails of the distribution.

kll_doubles(): KLL sketch for approximate quantiles of a numeric stream
kll_floats(): KLL sketch for approximate quantiles of a numeric stream stored as floats
req(): REQ sketch for relative-error approximate quantiles of a numeric stream
tdigest_double(): t-Digest sketch for approximate quantiles of a numeric stream

Cardinality sketches

HLL, CPC, and Theta sketches for approximate distinct counting. Theta sketches additionally support set operations.

hll(): HLL sketch for approximate distinct counting
cpc(): CPC sketch for approximate distinct counting
theta(): Theta sketch for approximate distinct counting and set operations
theta_union() theta_intersection() theta_difference() theta_jaccard(): Theta sketch set operations

Frequency sketches

Frequent Items and Count-Min sketches for approximate frequency estimation over a numeric or character stream.

frequent_items(): Frequent Items sketch for approximate frequency estimation
count_min(): Count-Min sketch for approximate point-frequency estimation
count_min_suggest_num_buckets() count_min_suggest_num_hashes(): Suggest Count-Min sketch parameters

Tuple sketches

Array of Doubles sketches, a Theta-extension that associates a fixed-size array of doubles with each retained key, for approximate distinct counting alongside per-column sums. Supports set operations.

array_of_doubles(): Array of Doubles (Tuple) sketch for estimating sums alongside distinct counts
array_of_doubles_union() array_of_doubles_intersection() array_of_doubles_difference(): Array of Doubles sketch set operations

Sampling sketches

VarOpt and EBPPS sketches sample weighted items from a stream: VarOpt for variance-optimal subset-sum estimation, and EBPPS as a modern alternative to reservoir sampling.

varopt(): VarOpt sketch for variance-optimal sampling and subset-sum estimation
varopt_union(): Combine two VarOpt sketches
ebpps(): EBPPS sketch for proportional-to-size sampling

Filters

Bloom filters for approximate set membership, sized up front by accuracy or by an explicit number of bits and hash functions.

bloom_filter(): Bloom filter for approximate set membership
bloom_filter_suggest_num_filter_bits() bloom_filter_suggest_num_hashes(): Suggest Bloom filter sizing parameters