sktime: python toolbox for time series- next-generation AI
summary
sktime extends its scikit‑learn–style API to load and run deep learning and foundation models (e.g. morai, tiny times mix, Times of M) and composes them with transformers and pipelines, letting a single fitted foundation model forecast many series. demos show fit/predict parity with classic forecasters, backing forecaster bootstrapping to produce probabilistic forecasts, and a marketplace for indexed models; some integrations are Linux‑only or still PRs and intervals need calibration but the composed workflows work end‑to‑end.
AI-written summary · about
chapters (20)
- 0:00Introduction to sktime
- 0:24What is sktime
- 1:53Composable architecture
- 4:22sKtime toolbox
- 5:26Foundation models for time series
- 6:28Foundation model challenges
- 8:16Foundation models in sktime
- 9:00Demo: Morai forecaster
- 10:48Composing Foundation models
- 11:51Pipeline example
- 12:33Probabilistic forecasts
- 13:40Foundation models under the hood
- 14:24Development and future
- 16:30State transitions
- 17:11API complexity
- 18:15Neutral model marketplace
- 19:22Growing ecosystem
- 20:54Open governance
- 22:40Community growth
- 23:45Get involved with sktime
related talks
01h 30msktime - python toolbox for time series
26mTime Series Analysis in Python: Easy Tools for Scientific Insight
01h 26mSktime—Python Toolbox for Time Series: How to Implement Your Own Estimator
01h 27mIntroduction to Machine Learning with Time Series
32mAdopting Open-Source Tools for Time Series Forecasting
34mTime series for scikit-learn people
19mForecasting for Everyone - Foundation Models and the Future of Time Series Prediction
27mBenchmarking Time Series Foundation Models with sktime