AI & DataBeginner
Dataset
PronunciationDAY-tuh-set
Definition
A collection of data used to train, test, or analyze something, such as a set of labeled images.
Where you hear it
Machine learning, analytics, and research.
Examples
We built a dataset of 500 scanned invoices to test the OCR.
The dataset is biased, so the model is unfair to some users.
Common mistake
Testing on the same data you trained on. Keep a separate test set.