Knowledge Maxxing← Back to topic library
ML MaxxingBeginner60 sec
Training set
The training set is the portion of a dataset actually used to fit a model's parameters during learning.
Prerequisite: Dataset
What matters
A model's parameters are adjusted only using the training set, so its performance on this data does not reflect how well it will generalize to new, unseen examples.
Example
If a dataset of 10,000 house sales is split, the 7,000 examples used to fit the model's parameters make up the training set.
Common misconception
Good performance on the training set alone does not indicate a good model; it may simply mean the model has memorized those specific examples.
Recall question
Why doesn't strong training-set performance alone prove a model is good?
Keywords and related concepts
training setmodel fittingdata splitvalidation set