Outlier
Also called: outliers, extreme observation, anomaly
An outlier is an observation far from the rest of the data. The decision that matters is whether it's an error or a signal: a fat-finger price is noise to be cleaned, while a crash day is the single most informative observation in a risk dataset. Deleting outliers because they inconvenience a model is the most common way to build something that works beautifully until the day it matters.
Where this is taught
Definitions are the trailer. These free levels turn Outlier into something you play — one bite-size lesson, with worked examples, a quiz and XP.