datasketch: Big Data Looks Small

https://github.com/ekzhu/datasketch/workflows/Python%20package/badge.svg https://zenodo.org/badge/DOI/10.5281/zenodo.290602.svg

datasketch gives you probabilistic data structures that can process and search very large amount of data super fast, with little loss of accuracy.

This package contains the following data sketches:

Data Sketch

Usage

MinHash

estimate Jaccard similarity and cardinality

Weighted MinHash

estimate weighted Jaccard similarity

HyperLogLog

estimate cardinality

HyperLogLog++

estimate cardinality

The following indexes for data sketches are provided to support sub-linear query time:

Index

For Data Sketch

Supported Query Type

MinHash LSH

MinHash, Weighted MinHash

Jaccard Threshold

MinHash LSH Forest

MinHash, Weighted MinHash

Jaccard Top-K

MinHash LSH Ensemble

MinHash

Containment Threshold

datasketch must be used with Python 2.7 or above and NumPy 1.11 or above. Scipy is optional, but with it the LSH initialization can be much faster.

Note that MinHash LSH and MinHash LSH Ensemble also support Redis and Cassandra storage layer (see MinHash LSH at Scale).

Install

To install datasketch using pip:

pip install datasketch

This will also install NumPy as dependency.

To install with Redis dependency:

pip install datasketch[redis]

To install with Cassandra dependency:

pip install datasketch[cassandra]

To install with Scipy for faster MinHashLSH initialization:

pip install datasketch[scipy]