Getting Scikit-Learn To Run On Top Of Pandas
summary
Claims that Scikit-Learn estimators become natively Pandas-aware by lazy dynamic wrapping at import time using a meta-path plus Python's inspect, converting DataFrame/Series to NumPy for the wrapped estimator and restoring indices/columns on output. It adds preserves DataFrame indices and column names safety checks, exposes pipe and plus operators for pipeline/concatenation syntax, and ships as a few hundred lines of adapter code that actually works across the Scikit-Learn ecosystem with minimal runtime overhead.
AI-written summary · about
chapters (8)
- 0:00Start
- 4:28Perform Non-Negative Matrix Factorization
- 15:14Implementation
- 15:29Wrapper Class
- 17:18Dynamic Loading
- 20:35Conclusion
- 22:26Questions
- 26:07What Exactly Is Pandas Impact on on Memory
related talks
35mPandas, Pipelines, and Custom Transformers
33mUsing Scikit-Learn's interface for turning Spaghetti Data Science into Maintainable Software
27mSoftware Library APIs: Lessons Learned from scikit-learn
35mScikit-Learn can do THAT?!
35mIdiomatic Pandas
42mWriting a scikit-learn compatible estimator in the modern age
01h 13mFrom Pandas to PySpark
28mIdiomatic Pandas