AI 101

Dataset

Also known as: datasets · data

Basics

A dataset is a structured collection of examples. In machine learning it’s the raw material the model learns from (or is tested on).

  • Training set — what the model learns from.
  • Validation set — for tuning while training.
  • Test set — the held-out report card.

The quality and breadth of the data often matters more than the architecture.

Golden rule: "garbage in, garbage out". A smarter model on bad data usually loses to a simpler model on great data.