Sktime
sktime is the open-source Python framework for time series machine learning — 500+ models behind one scikit-learn-style fit/predict API.
sktime is the broadest free Python option when your problem is more than plain forecasting — classification, clustering, anomaly detection, and time-to-event modeling all share the same fit/predict grammar. Pick it when you want to benchmark classical baselines against foundation models like Amazon's Chronos without rewriting pipelines. Between the 1.0 series and the 1.2.0 changelog this year, the interface is still moving, so budget time for version pinning.
Verified 3d ago · liveness 70/100 · cite: rightaichoice.com/tools/sktime
- Python data scientists prototyping forecasting, classification, and anomaly detection in one framework
- Researchers benchmarking classical statistical models against AI foundation models like Chronos
- Developers who want to swap models with a two-line edit and keep composable pipelines intact
- Teams that need a permissive BSD-3-Clause license and no vendor lock-in on their data
- Teams needing distributed computing for large-scale or terabyte-sized time series data
- Non-programmers who want a graphical interface or no-code forecasting environment
- Workflows that must operate outside the Python ecosystem or via standalone CLI batch jobs
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip sktime if you need distributed training on terabyte-scale series, a no-code or GUI interface, or a CLI batch workflow — this is an in-memory Python library, not a platform.
The library itself is free under BSD-3-Clause, so the real cost comparison is your engineering time: sktime costs nothing against paid forecasting platforms but expects Python skills and version discipline. Darts and Prophet are comparably free with narrower scope. Teams wanting an SLA, forward-deployed engineers, and training pay the vendor's custom enterprise rate, which is a services relationship rather than a software subscription.
In short
Sktime — sktime is the open-source Python framework for time series machine learning — 500+ models behind one scikit-learn-style fit/predict API. Best for Python data scientists prototyping forecasting, classification, and anomaly detection in one framework, Researchers benchmarking classical statistical models against AI foundation models like Chronos, Developers who want to swap models with a two-line edit and keep composable pipelines intact. Free to use.
What's new in Sktime
Checked 2 days agoAcross the latest 6 updates: 2 launches, 3 changelog entries and 1 community discussion.
TServe: self-host every forecasting model
TServe serves Chronos-2, TimesFM, Moirai and 100+ forecasting checkpoints, plus custom sktime models, from a self-hosted server.
sktime 1.2.0 released
sktime 1.2.0 ships dependency changes, core interface changes, deprecations and removals alongside enhancements and fixes.
The Temporer Has No Clothes
sktime argues gating forecasting foundation models is artificial lock-in, noting state-of-art models can run directly in a browser.
sktime is 1.0
sktime 1.0 adds foundation models, pretraining, a reusable benchmarking framework and agentic forecasting behind one fit/predict interface.
sktime 1.1.0 released
sktime 1.1.0 covers deprecations and removals following the 1.0 release.
sktime 1.0.2 released
sktime 1.0.2 patch adds enhancements, documentation updates, maintenance and fixes.
What people actually say about Sktime — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
6 mentions across 2 sources (Hacker News, GitHub) · researched Jul 3, 2026.
Average across the 2 sources that answered — each source counts once, not each post.
- +Unified scikit-learn-like API across forecasting, classification, regression, and clustering.
- +Rich composable pipelines and ensembles for complex model building.
- +Active community with regular updates and innovative features like Craft().
- +Free and permissive BSD-3-Clause license.
- +Handles medium-sized datasets well with pandas/NumPy.
- −Over 2,300 open GitHub issues suggests maintenance strain.
- −Documentation for advanced features can be sparse.
- −Lacks native categorical feature support for classification/forecasting.
- −LLM-driven features are experimental, not production-ready.
- −Occasionally omitted from comparisons with similar libraries like aeon.
- • No official commercial support; relies on community
- • May require additional libraries for production deployments
Viability Score
How well maintained and how widely used is Sktime? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Unified scikit-learn-style API with fit/predict/transform across all supported tasks
- 500+ estimators spanning classical statistics, machine learning, deep learning, and AI foundation models
- Foundation model support including Chronos via ChronosForecaster (amazon/chronos-bolt-tiny)
- Agentic forecasting added in the 1.0 release series
- Pretraining support introduced with the 1.0 release
- Reusable benchmarking framework for comparing models on your own series
- Probabilistic forecasting with confidence and uncertainty ranges instead of single-point values
- Time series classification for signal types, product groups, and machine states
- Time series regression for targets such as remaining equipment life
- Time series clustering to surface segments, product families, and hidden patterns
- Anomaly and change point detection on sensor, fraud, and demand data
- Feature extraction turning raw series into model-ready inputs
- Time-to-event (survival) modeling for churn, failure, and conversion timing
- Time series alignment for sequences at different speeds or start times
- Composable pipelines with transformations, ensembles, and hyperparameter tuning
About Sktime
sktime is a free, open-source Python library (BSD-3-Clause) that covers the time series machine learning lifecycle through a single scikit-learn-style interface. Install it with pip install sktime, then use the same fit/predict/transform pattern for forecasting, classification, regression, clustering, anomaly and change detection, feature extraction, probabilistic prediction, time-to-event modeling, and alignment. The practical draw is model breadth under one harness: 500+ estimators spanning classical statistics, machine learning, deep learning, and AI foundation models. Swapping an AutoARIMA baseline for a foundation model is a two-line edit — your data loading, fit, and predict calls do not change. The docs demo this directly with ChronosForecaster("amazon/chronos-bolt-tiny") from sktime.forecasting.chronos. You can also register your own estimators, transformers, and adapters from a closed codebase and drive them through the same API, which matters if you need private components next to public ones. Version 1.0 landed in June 2026, and the changelog has moved fast since: 1.1.0 in July 2026, then 1.2.0 in September 2026, each shipping dependency changes, core interface changes, deprecations and removals. The 1.0 series added foundation models, pretraining, a reusable benchmarking framework, and agentic forecasting. This is a codebase actively consolidating interfaces, so pinning versions is not optional. Who it is for: Python data scientists, researchers, and developers building forecasting or classification pipelines on in-memory, medium-sized datasets. Retail demand planning and energy load forecasting are the two flagship use cases the project pushes. It is community-governed, with a reported 10.1k GitHub stars, 532+ contributors, and roughly 1.1M monthly downloads. Enterprise engagements exist for teams wanting implementation help, an SLA, and forward-deployed engineers.
Behind the Verdict
Where sktime earns its place is scope. Most Python forecasting libraries do forecasting, full stop. sktime covers forecasting, classification, regression, clustering, anomaly and change detection, feature extraction, probabilistic prediction, time-to-event modeling, and alignment behind one fit/predict/transform API — so a team that starts with demand forecasting and later needs anomaly detection on the same sensor feeds does not shop for a second library. In practice the two-line model swap is the feature people actually stay for. You fit AutoARIMA, you fit ChronosForecaster("amazon/chronos-bolt-tiny"), and the surrounding pipeline — transformations, ensembles, hyperparameter tuning — survives both. That makes it a reasonable place to run cheap benchmarks of classical statistics against modern foundation models before committing budget to either. We would reach for it when the data fits in memory, the team writes Python, and someone is comfortable reading changelogs. Retail demand planning and energy load forecasting are the documented sweet spots, and the probabilistic output — ranges rather than point estimates — is what planners actually need to weigh stockout risk against carrying cost. Where it bites: this is not a distributed engine. Terabyte-scale series, cluster-scale training, and streaming infrastructure are outside what the sources describe, and there is no graphical interface for analysts who do not code. If your organization wants a dashboard-first forecasting product or a vendor SLA on day one, a commercial platform is the shorter path. The other caveat is churn. Three changelog entries in four months — 1.0.0 in June 2026, 1.1.0 in July, 1.2.0 in September — each with deprecations and removals, is a sign of healthy consolidation but also a sign you
Researching Sktime? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Sktime actually fits — and what changes day-one when you adopt it.
You load weekly SKU-level sales into a pandas DataFrame, fit AutoARIMA(sp=12) as a baseline, then swap in ChronosForecaster("amazon/chronos-bolt-tiny") by editing two lines to see whether the foundation model beats the classical baseline.
Outcome: You get probabilistic demand forecasts with uncertainty ranges for stocking decisions, and a defensible benchmark showing which model wins on your data — without rewriting any load, fit, or predict code.
You pull hourly load data, build a pipeline that combines feature extraction with a regression estimator, and use the benchmarking framework to compare several models across your own series.
Outcome: You identify the evening peak with calibrated confidence ranges, so the team can size generation to risk instead of over-provisioning every hour.
You feed vibration sensor series into sktime's anomaly and change point detection, then run time-to-event modeling on the same data to estimate failure timing.
Outcome: You get an early failure-risk signal and a rough time horizon, both built on the same API you already used for classification.
Use Cases
- Forecast product demand across products, stores, and regions for retail planning
- Forecast hourly energy load to balance the grid and cut peak costs
- Classify sensor signals for predictive maintenance or activity recognition
- Cluster customer time series to identify purchasing behavior segments
- Detect anomalies in energy load data before equipment failure
- Model time-to-event outcomes like equipment failure or customer churn
- Build probabilistic forecasts with confidence ranges for volatility and risk modeling
- Benchmark classical statistical baselines against foundation models like Chronos
Models Under the Hood
as of 2026-10-17
Limitations
- sktime performs in-memory computation on a single machine with no distributed computing, and relies on pandas and NumPy containers, which limits scalability to medium-sized data.
- It runs via an interactive Python interpreter and provides no command-line interface or graphical user interface.
- Version 1.2.0 and the preceding 1.0 series shipped dependency and core interface changes plus deprecations and removals, so upgrading without pinning can break working pipelines.
- It is community-governed under BSD-3-Clause, with support through GitHub unless you engage the paid enterprise offering.
as of 2026-10-07
Verification history
We have re-verified Sktime 10 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 10 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Sktime tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source
$0
Ideal for
Python data scientists, researchers, and developers building forecasting or classification pipelines in-house on medium-sized in-memory datasets
What this tier adds
Starting tier: free under BSD-3-Clause with the full 500+ estimator library, unified API, composable pipelines, and community support via GitHub
Enterprise
Custom
Ideal for
Organizations that want sktime implemented and operated inside their company with production support and forward-deployed engineering help
What this tier adds
Adds on top of open source: implementation inside your company, production-grade SLA and support, forward-deployed engineers, and training and consulting
Where the pricing makes sense
The company stage and team size where Sktime's pricing actually pencils out — and where peers do it cheaper.
The library itself is free under BSD-3-Clause, so the real cost comparison is your engineering time: sktime costs nothing against paid forecasting platforms but expects Python skills and version discipline. Darts and Prophet are comparably free with narrower scope. Teams wanting an SLA, forward-deployed engineers, and training pay the vendor's custom enterprise rate, which is a services relationship rather than a software subscription.
Setup time & first value
How long it actually takes to get something useful out of Sktime — broken out by persona, not the marketing-page minute.
Python developers reach a first forecast in minutes: `pip install sktime`, load a series with `load_airline()`, fit a forecaster, and predict. The getting-started guide covers core time series concepts for newcomers. Going from a single model to a tuned, composed pipeline across a real dataset takes days, not hours.
Switching to or from Sktime
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Prophet: Replace the Prophet fit/predict calls with an sktime forecaster inside the same sklearn-style fit/predict pattern, then benchmark the two on your own series.
- →From Darts: Re-implement your Darts pipelines using sktime's composable transformations and ensembles, both expose sklearn-style APIs so the mapping is mostly mechanical.
- →From a commercial forecasting platform: Export your series to pandas, rebuild the pipeline in sktime, and use the benchmarking framework to check you haven't lost accuracy.
- ↗To Darts: Rebuild pipelines in Darts' own model API if you want a narrower, slower-moving surface than sktime's 1.0-series interface consolidation.
- ↗To a commercial platform: Move to a hosted forecasting service if you need distributed training, an SLA, or a no-code interface rather than an in-memory Python library.
Resources & Guides
- Resourcesktime.net
Stable · Sktime
Helpful link from sktime.net
- Documentationsktime.net
Docs · Sktime
Full product docs from sktime.net
- Resourcesktime.net
Get Started · Sktime
Helpful link from sktime.net
- Resourcesktime.net
Api Reference · Sktime
Helpful link from sktime.net
- Resourcesktime.net
Tutorials · Sktime
Helpful link from sktime.net
- Resourcesktime.net
Changelog · Sktime
Helpful link from sktime.net
- Resourcesktime.net
Estimator Overview · Sktime
Helpful link from sktime.net
- Resourcegithub.com
Sktime · Sktime
Helpful link from github.com
Tutorials & Learning
YouTube returned 6 videos for “Sktime”, and we withheld 6: 6 could not be judged, because “Sktime” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Sktime.
Official links
Tools that pair well with Sktime
Common stack mates teams adopt alongside Sktime, with the specific reason each pairing earns its keep.
Obviously AI
Obviously AI is now Zams — the archived obviously.ai domain shows the old no-code machine learning workflow: CSV in, predictive model out,
Chat2DB
Chat2DB turns plain English into SQL across 40+ databases in an open-source AI SQL client that keeps query execution on your own machine.
OpenAgents
OpenAgents is an open-source language agent platform for data analysis, 200+ plugins and autonomous web browsing.
Featured Head-to-Head Comparisons
Sktime vs Screenplayiq
ScreenplayIQ and Sktime serve entirely different needs, so the choice is clear based on your domain: screenwriters seeking data-driven script feedback should choose ScreenplayIQ, while data scientists doing time series ML should pick Sktime. There is no overlap in use cases.
Sktime vs Praktika
Praktika and Sktime serve completely different domains: Praktika is a mobile-first app for conversational language practice with AI tutors, while Sktime is a Python library for time series machine learning. Choose Praktika if you want to improve speaking fluency through interactive, feedback-driven conversations on your phone. Choose Sktime if you are a data scientist or ML engineer needing a unified, scikit-learn‑like framework for forecasting, classification, or regression on time series data.
Alternatives to Sktime
View allObviously AI
Obviously AI is now Zams — the archived obviously.ai domain shows the old no-code machine learning workflow: CSV in, predictive model out,
Chat2DB
Chat2DB turns plain English into SQL across 40+ databases in an open-source AI SQL client that keeps query execution on your own machine.
OpenAgents
OpenAgents is an open-source language agent platform for data analysis, 200+ plugins and autonomous web browsing.
Frequently Asked Questions
Categories
Best-of guides
Used Sktime? Help shape our editorial sentiment research.