Lecture 5: Preprocessing and sklearn pipelines

Preprocessing and sklearn pipelines

Slides

Download slides as PDF

View slides in full screen

Outline

  • The full ML workflow and why data representation matters
  • Restaurant survey demo: imputation, scaling, and categorical encoding
  • Transformers and predictive estimators in sklearn
  • Combining preprocessing and prediction in a pipeline
  • The golden rule in preprocessing and cross-validation
  • Motivation for ColumnTransformer