A tool for detecting anomalies in time series data


Keywords
anomaly, detection, timeseries, model-free, adaptive
License
MIT-feh
Install
pip install patternly==0.0.33

Documentation

patternly

patternly logo

Info: Paper draft link will be posted here
Author: Drew Vlasnik, Ishanu Chattopadhyay
Laboratory: The Laboratory for Zero Knowledge Discovery, The University of Chicago https://zed.uchicago.edu
Description: Discovery of emergent anomalies in data streams without explicit prior models of correct or aberrant behavior, based on the modeling of ergodic, quasi-stationary finite valued processes as probabilistic finite state automata (PFSA).
Documentation: https://zeroknowledgediscovery.github.io/patternly

Installation:

pip install patternly --user -U

Usage:

See examples.

from patternly.detection import AnomalyDetection, StreamingDetection