sktime
sktimePython library with one scikit-learn-style interface across time-series learning tasks - forecasting, classification, regression, clustering, anomaly and changepoint detection - and tools for pipelining, ensembling, tuning and reduction between them. It interfaces statsmodels, Prophet, PyOD and other libraries alongside its own estimators. A consensus-governed community project, with support services sold by a separate company.
Openness
5 high confidence- license
- BSD-3-Clause(OSI)
- source
- public(github.com/sktime/sktime)
- core features withheld
- no — support and custom work sold by sktime Enterprise Solutions GmbH
BSD-3-Clause, and the library builds whole from the public repository under an elected community council. The enterprise offering is custom model development, integration and SLA-backed support from a separate company, not a withheld edition of the library.
- https://cdn.jsdelivr.net/gh/sktime/sktime@main/LICENSE recorded 2026-09-26
LICENSE: "BSD 3-Clause License", "Copyright (c) 2019 - present, The sktime developers."
- https://ungh.cc/repos/sktime/sktime/files/main recorded 2026-09-26
Repository tree of 2,219 files with no ee, enterprise, commercial or pro directory.
- https://www.sktime.net/about/ recorded 2026-09-26
About page, on the enterprise team: "Professional support when you take sktime to production ... A separate company that keeps the project open".
- https://www.sktime.net/en/stable/get_involved/governance.html recorded 2026-09-26
Governance: "sktime is a consensus-based community project." and "The community council is elected by contributors to sktime in regular elections."
- https://www.sktime.net/enterprise/ recorded 2026-09-26
Enterprise page: "Custom solutions, seamless integration, and dedicated support from the core team behind sktime", offering custom model development, infrastructure integration and SLA-backed support.
Adoption
4 high confidenceMeasured on monthly PyPI downloads of the sktime package.
- https://pypistats.org/api/packages/sktime/recent recorded 2026-09-26
last_month 1,240,731 downloads of sktime
Capability
3 high confidencesktime does for time series what scikit-learn does for tables: many learners behind one interface, with the composition and tuning tools to compare them. Picking and combining models is still the user's job.
- https://cdn.jsdelivr.net/gh/sktime/sktime@main/README.md recorded 2026-09-26
README: "It provides a unified interface for multiple time series learning tasks. Currently, this includes forecasting, time series classification, clustering, anomaly/changepoint detection, and other tasks." and "tools for composite model building, such as pipelining, ensembling, tuning, and reduction".
- https://www.sktime.net/en/stable/ recorded 2026-09-26
Docs landing: "Browse all 500+ models"; "A unified Python API for the complete time series machine learning lifecycle".
Verified 2026-09-26