Streaming Algorithms Python Packages

pysuricata

Single-pass exploratory data analysis on streaming algorithms. Bounded memory at any dataset size, pandas and Polars, self-contained HTML reports.

aberrant

Online Anomaly Detection Algorithms

npstreams

Streaming operations on NumPy arrays

llm-sketchkit

Count distinct LLM users and find top token consumers across services, days and tools. Fast, small, portable summaries in Go and Python.

pyonlinesvr

Python-Wrapper for Francesco Parrella's OnlineSVR C++ implementation with scikit-learn-compatible interface.

onad

Online Anomaly Detection Algorithms

sqsketch

Fixed-size sketches of probability distributions: Bhattacharyya coefficient and Hellinger distance from ~1 KB per profile, with error independent of alphabet size. Library + fully verified paper (every number traced to a committed script).

cardinalitykit

A comprehensive Python toolkit for cardinality estimation algorithms (Flajolet-Martin, LogLog, SuperLogLog, HyperLogLog, HyperReal) — enabling privacy-preserving analytics, audience measurement, and sample-to-population extrapolation through probabilistic counting techniques.

kll-sketch

High-integrity Python KLL sketch for streaming quantiles — mergeable, deterministic, zero-dependency, with optional C++17/SIMD acceleration.

randomhash

A simple, time-tested, family of random hash functions in Python, based on CRC32 and xxHash, affine transformations, and the Mersenne Twister. 🎲

sublinear-algorithms

Python library implementing a subset of streaming algorithms. Includes variations of these algorithms (e.g. adversarially robust), as well as support for multiple data types.

rivertext

RiverText is a framework that standardizes the Incremental Word Embeddings proposed in the state-of-art. Please feel welcome to open an issue in case you have any questions or a pull request if you want to contribute to the project!

sublinear

Python library implementing a subset of streaming algorithms. Includes variations of these algorithms (e.g. adversarially robust), as well as support for multiple data types.