# khundman/telemanom

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.

Repository: https://github.com/khundman/telemanom
Canonical: https://ross.abutalabs.com/products/telemanom
Homepage: https://arxiv.org/abs/1802.04431
Language: Jupyter Notebook
License: NOASSERTION
License Family: other
Topics: deep-learning, lstm, time-series, keras, tensorflow, kdd2018, kdd, rnn, anomaly-detection
Last push: 2025-01-17T13:42:45+00:00

## Health v2 (maintenance only)
Score: 33/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 2, release rhythm 35, longevity 100
- inputs: {"age_days": 3015, "days_push": 593, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1187, forks 267 (observed 2026-08-28T04:03:55.227154+00:00)

## What it is
Telemanom is a Python framework that uses vanilla LSTMs (Keras/TensorFlow) to detect anomalies in multivariate time series sensor data, with nonparametric dynamic thresholding of prediction errors. It includes the spacecraft telemetry datasets and experiments from the NASA SMAP and Mars Science Laboratory missions from a 2018 KDD paper.

## Use cases
- detect anomalies in multivariate sensor telemetry
- reproduce the KDD 2018 spacecraft anomaly detection experiments
- train LSTM models on normal system behavior and flag deviations
- benchmark anomaly detection on the SMAP and MSL labeled datasets
- apply unsupervised dynamic thresholding to time series prediction errors
- monitor IoT or industrial sensor streams for unexpected behavior

## When to choose
- you need LSTM-based anomaly detection for multivariate time series
- you want a research-proven baseline with labeled spacecraft anomaly data
- you need unsupervised, nonparametric error thresholding without labeled anomalies

## When to avoid
- you need real-time streaming detection at scale in production
- you want transformer or modern architecture approaches
- you need actively maintained tooling with current TensorFlow versions

## Facets
- artifact type: framework
- maturity: maintenance
- function: machine-learning, deep-learning, data-science, analytics
- domain: machine-learning, time-series, data-science, artificial-intelligence
- platform: python, cross-platform
- tags: lstm, anomaly-detection, time-series, keras, tensorflow, telemetry, nasa, kdd2018, sensor-data, docker

## Member repositories
- khundman/telemanom (main) score 33

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:55.227154+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:23:42.642661+00:00, confidence not recorded.
  - readme: https://github.com/khundman/telemanom (fetched 2026-08-28T04:03:55.227154+00:00, sha 4b464b37f3cf)
  - homepage: https://arxiv.org/abs/1802.04431 (fetched 2026-08-29T12:31:00.566026+00:00, sha f73b21ce2d5c)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T12:31:00.572187+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T12:31:00.576511+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T12:31:00.578781+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T12:31:00.574621+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.