awesome-TS-anomaly-detection

awesome-TS-anomaly-detection

List of tools & datasets for anomaly detection on time-series data.

All lists are in alphabetical order. In the lists, maintaned projects are prioritized vs not mantained. A repository is considered “not maintained” if the latest commit is > 1 year old, or explicitly mentioned by the authors.

🎓 If you found this collection useful and want to cite it, please follow the citation policy.

Anomaly Detection Software

Name	Language	Pitch	License	Maintained
Cuebook’s CueObserve	Python3	Anomaly detection on SQL data warehouses and databases.	Apache-2.0	:heavy_check_mark:
Yahoo’s EGADS	Java	GADS is a library that contains a number of anomaly detection techniques applicable to many use-cases in a single package with the only dependency being Java.	GPL	:heavy_check_mark:
AIStream’s flow-forecast	Python	Deep learning PyTorch library for time series forecasting, classification, and anomaly detection (originally for flood forecasting).	GPL-3	:heavy_check_mark:
Hastic	Python + node.js	Anomaly detection tool for time series data with Grafana-based UI.	GPL	:heavy_check_mark:
Zillow’s Luminaire	Python	Luminaire is a python package that provides ML driven anomaly detection and forecasting solutions for time series data.	Apache-2.0	:heavy_check_mark:
MIDAS	C++	MIDAS, short for Microcluster-Based Detector of Anomalies in Edge Streams, detects microcluster anomalies from an edge stream in constant time and memory.	Apache-2.0	:heavy_check_mark:
Orion	Python	Orion is a machine learning library built for unsupervised time series anomaly detection, providing a number of “verified” ML pipelines (a.k.a Orion pipelines) that identify rare patterns and flag them for expert review.	MIT	:heavy_check_mark:
OutlierDetection.jl	Julia	Fast, scalable and flexible Outlier Detection with Julia.	MIT	:heavy_check_mark:
PyOD	Python	PyOD is a comprehensive and scalable Python toolkit for detecting outlying objects in multivariate data.	BSD 2-Clause	:heavy_check_mark:
ruptures	Python	Ruptures is a Python library for off-line change point detection. This package provides methods for the analysis and segmentation of non-stationary signals.	BSD 2-Clause	:heavy_check_mark:
EarthGecko Skyline	Python3	Skyline is a real-time anomaly detection system, built to enable passive monitoring of hundreds of thousands of metrics.	MIT	:heavy_check_mark:
Expedia.com’s Adaptive Alerting	Java	Streaming anomaly detection with automated model selection and fitting.	Apache-2.0	❌
Arundo’s ADTK	Python	Anomaly Detection Toolkit (ADTK) is a Python package for unsupervised / rule-based time series anomaly detection.	MPL 2.0	❌
Twitter’s AnomalyDetection	R	AnomalyDetection is an open-source R package to detect anomalies which is robust, from a statistical standpoint, in the presence of seasonality and an underlying trend.	GPL	❌
Lytics’ Anomalyzer	Go	Anomalyzer implements a suite of statistical tests that yield the probability that a given set of numeric input, typically a time series, contains anomalous behavior.	Apache-2.0	❌
banpei	Python	Outlier detection (Hotelling’s theory) and Change point detection (Singular spectrum transformation) for time-series.	MIT	❌
Ele.me’s banshee	Go	Anomalies detection system for periodic metrics.	MIT	❌
CAD	Python	Contextual Anomaly Detection for real-time AD on streagming data (winner algorithm of the 2016 NAB competition).	AGPL	❌
Chaos Genius	Python	ML powered analytics engine for outlier/anomaly detection and root cause analysis.	MIT	❌
Mentat’s datastream.io	Python	An open-source framework for real-time anomaly detection using Python, Elasticsearch and Kibana.	Apache-2.0	❌
DeepADoTS	Python	Implementation and evaluation of 7 deep learning-based techniques for Anomaly Detection on Time-Series data.	MIT	❌
Donut	Python	Donut is an unsupervised anomaly detection algorithm for seasonal KPIs, based on Variational Autoencoders.	-	❌
LoudML	Python	Loud ML is an open source time series inference engine built on top of TensorFlow. It’s useful to forecast data, detect outliers, and automate your process using future knowledge.	MIT	❌
Linkedin’s luminol	Python	Luminol is a light weight python library for time series data analysis. The two major functionalities it supports are anomaly detection and correlation. It can be used to investigate possible causes of anomaly.	Apache-2.0	❌
Numenta’s Nupic	C++	Numenta Platform for Intelligent Computing is an implementation of Hierarchical Temporal Memory (HTM).	AGPL	❌
oddstream	R	oddstream (Outlier Detection in Data Streams) provides real time support for early detection of anomalous series within a large collection of streaming time series data.	GPL-3	❌
PyOdds	Python	PyODDS is an end-to end Python system for outlier detection with database support. PyODDS provides outlier detection algorithms, which support both static and time-series data.	MIT	❌
PySAD	Python	PySAD is a streaming anomaly detection framework with various online models and complete set of tools for experimentation.	BSD 3-Clause	❌
rrcf	Python	Implementation of the Robust Random Cut Forest algorithm for anomaly detection on streams.	MIT	❌
Netflix’s Surus	Java	Robust Anomaly Detection (RAD) - An implementation of the Robust PCA.	Apache-2.0	❌
NASA’s Telemanom	Python	A framework for using LSTMs to detect anomalies in multivariate time series data. Includes spacecraft anomaly data and experiments from the Mars Science Laboratory and SMAP missions.	custom	❌

Related Software

This section includes some time-series software for anomaly detection-related tasks, such as forecasting, generic TS analysis and labeling.

Forecasting

Name	Language	Pitch	License	Maintained
darts	Python	darts is a python library for easy manipulation and forecasting of time series. It contains a variety of models, from classics such as ARIMA to neural networks.	Apache-2.0	:heavy_check_mark:
ETNA	Python	etna is a python library for time series forecasting and analysis with temporal data structure always in mind. Includes a variety of predictive models with unified interface along with EDA and validation methods.	Apache-2.0	:heavy_check_mark:
Amazon’s GluonTS	Python	GluonTS is a Python toolkit for probabilistic time series modeling, built around MXNet. GluonTS provides utilities for loading and iterating over time series datasets, state of the art models ready to be trained, and building blocks to define your own models.	Apache-2.0	:heavy_check_mark:
pmdarima	Python	Porting of R’s auto.arima with a scikit-learn-friendly interface.	MIT	:heavy_check_mark:
Facebook’s Prophet	Python/R	Prophet is a procedure for forecasting time series data. It is based on an additive model where non-linear trends are fit with yearly and weekly seasonality, plus holidays.	BSD	:heavy_check_mark:
PyFlux	Python	The library has a good array of modern time series models, as well as a flexible array of inference options (frequentist and Bayesian) that can be applied to these models.	BSD 3-Clause	❌

Time-Series Analysis

Name	Language	Pitch	License	Maintained
Facebook’s Kats	Python	Kats aims to provide the one-stop shop for time series analysis, including detection, forecasting, feature extraction/embedding, multivariate analysis, etc.	MIT	:heavy_check_mark:
MatrixProfile	Python	A Python 3 library making time series data mining tasks, utilizing matrix profile algorithms, accessible to everyone.	Apache-2.0	:heavy_check_mark:
Salesforce’s Merlion	Python	Merlion: A Machine Learning Framework for Time Series Intelligence, support multiple models including ARIMA	BSD 3-Clause	:heavy_check_mark:
SaxPy	Python	General implementation of SAX, as well as HOTSAX for anomaly detection.	GPLv2.0	:heavy_check_mark:
sktime	Python	A unified framework for machine learning with time series. It provides specialized time series algorithms and scikit-learn compatible tools to build, tune and validate time series models for multiple learning problems.	BSD 3-Clause	:heavy_check_mark:
sktime-dl	Python	An extension package for deep learning with Tensorflow/Keras for sktime.	BSD 3-Clause	:heavy_check_mark:
Squey	C++	Visualization software allowing in-depth exploration of timeseries while displaying every outliers.	MIT	:heavy_check_mark:
Tigramite	Python	Tigramite is a causal time series analysis python package. It allows to efficiently reconstruct causal graphs from high-dimensional time series datasets and model the obtained causal dependencies for causal mediation and prediction analyses.	GPLv3.0	:heavy_check_mark:
tsflex	Python	tsflex is a time series toolkit for feature extraction & processing that is both flexible and efficient. This package supports strided-window feature extraction on multivariate, irregularly-sampled sequence data.	MIT	:heavy_check_mark:
tslearn	Python	tslearn is a Python package that provides machine learning tools for the analysis of time series. This package builds on scikit-learn, numpy and scipy libraries.	BSD 2-Clause	:heavy_check_mark:
seglearn	Python	Seglearn is a python package for machine learning time series or sequences. It provides an integrated pipeline for segmentation, feature extraction, feature processing, and final estimator.	BSD 3-Clause	❌

Labeling

Name	Language	Pitch	License	Maintained
Baidu’s Curve	Python	Curve is an open-source tool to help label anomalies on time-series data.	Apache-2.0	❌
Microsoft’s Taganomaly	R (dockerized web app)	Simple tool for tagging time series data. Works for univariate and multivariate data, provides a reference anomaly prediction using Twitter’s AnomalyDetection package.	MIT	❌

Benchmark Datasets

• Numenta’s NAB

NAB is a novel benchmark for evaluating algorithms for anomaly detection in streaming, real-time applications. It is comprised of over 50 labeled real-world and artificial timeseries data files plus a novel scoring mechanism designed for real-time applications.

• Yahoo’s Webscope S5

The dataset consists of real and synthetic time-series with tagged anomaly points. The dataset tests the detection accuracy of various anomaly-types including outliers and change-points.

• 2020 Skoltech’s SKAB

SKAB (Skoltech Anomaly Benchmark) is designed for evaluating algorithms for anomaly detection. The benchmark currently includes 30+ datasets plus Python modules for algorithms’ evaluation. Each dataset represents a multivariate time series collected from the sensors installed on the testbed. All instances are labeled for evaluating the results of solving outlier detection and changepoint detection problems.

• 2018 AIOps’s KPI-Anomaly-Detection

The dataset consists of KPIs (key performace index) time series data from many real scenarios of Internet companies with ground truth label. Click below for a more detailed description.

Details

The dataset consists of KPI (key performace index) time series data from many real scenarios of Internet companies with ground truth label. KPIs fall into two broad categories: service KPIs and machine KPIs. Service KPIs are performance metrics that reflect the size and quality of a Web service, such as page response time, page views, and number of connection errors. Machine KPIs are performance indicators that reflect the health of the machine (server, router, switch), such as CPU utilization, memory utilization, disk IO, network card throughput, etc.

Dataset Descriptions:

In order to train the anomaly detection algorithm, the training KPI data provided by us is shown in table 1, including four columns: KPI ID, timestamp, the value of the KPI at that time, and whether the time is abnormal (label).

Table 1 training KPI data case

KPI ID	Timestamp	Value	Label
0	1503831000	10.8	0
0	1503831060	12.3	1
…	…	…	…

In order to evaluate the anomaly detection algorithm, the test KPI data provided by us is shown in table 2, including two columns: KPI ID, timestamp, and the value corresponding to the KPI at that time.

Table 2 testing KPI data case

KPI ID	Timestamp	Value
0	1503831000	10.8
0	1503831060	12.3
…	…	…

Citation Policy

If you would like to cite this repository, please use the following DOI:

or use this BibTex

 @article{medico2020, 
          title={rob-med/awesome-TS-anomaly-detection}, 
          DOI={10.5281/zenodo.3972944}, 
          abstractNote={A collection of tools and datasets for anomaly detection on time-series data.}, 
          publisher={Zenodo}, author={Roberto Medico}, year={2020}, month={Aug}}