Cross-Validation
Holding out a subset of data from fitting and scoring the fitted model on it, to estimate generalisation rather than memorisation. The held-out score rises when extra parameters are fitting noise.
This Concept is waiting for its first lesson!
Holding out a subset of data from fitting and scoring the fitted model on it, to estimate generalisation rather than memorisation. The held-out score rises when extra parameters are fitting noise.
Are you a teacher? Sign in to start contributing.
Sign In