a workbench for creating scikit-learn like parametric objects and libraries
summary
skbase is a working meta‑toolkit that makes it easy to build scikit‑learn‑style libraries: it provides BaseObject/BaseEstimator classes, a tag and config system that inherits through class hierarchies, pipeline/composite helpers that chain fit/transform/predict, a discovery utility (all_objects discovery) and templateable tests with pytest fixtures. The codebase is used to rework sktime patterns, reduces extender boilerplate, and demonstrably runs the example pipelines and test harness in the repository.
AI-written summary · about
chapters (13)
- 0:00SK time and SK bass intro
- 2:10Simple SK bass example with permutations
- 8:26Tags and config in SK bass
- 12:00Machine learning pipelines with SK bass
- 15:59Fit methods and base estimators
- 18:58Testing utilities in SK bass
- 20:46Building SK bass package
- 25:28SK-learn interfaces and contracts
- 30:27Base classes and template patterns
- 44:40Composite estimators and pipelines
- 50:24Lookup and search functionality with all objects
- 53:56Testing framework in SK bass
- 57:27Contributing to SK time and SK bass
related talks
01h 30msktime - python toolbox for time series
01h 26mSktime—Python Toolbox for Time Series: How to Implement Your Own Estimator
30msktime: python toolbox for time series- next-generation AI
27mSoftware Library APIs: Lessons Learned from scikit-learn
37mScikit build: A build
42mWriting a scikit-learn compatible estimator in the modern age
26mImproving production workflows for scikit-learn models with skops
35mScikit-Learn can do THAT?!