Keep the rare sensor failures visible

A journey through when the important class is rare with Botie.

Data Preparation · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

Learning From a Class That Appears Only Occasionally

Part 1 of 2

Classification datasets often contain many examples of one outcome and few of another. Researchers found that a method could favour the majority while performing poorly on the minority that mattered to the application.

Sources for this history

Chawla and colleagues: SMOTE (2002)

← Previous lesson: What Could We Know at Prediction Time?Read the full lesson and derivations