306Course · Practitioner
Classification
The other half of supervised learning.
The other half of supervised learning. Logistic regression, the ROC/PR family, calibration, and the alternatives — with an operating-point choice defended in writing.
- 01
Fit a logistic regression.
- 02
Compare classifiers.
- 03
Read a confusion matrix.
- 04
Choose an operating point defensibly.
Prerequisites
- — Regression.
Pacing
~4 hours.
- IModule
Logistic regression
The starting point.
- IIModule
Trees and ensembles
The tabular workhorses.
A Note from the Faculty
Calibration is the property that makes predicted probabilities usable.