Dependent Variable

Also known as: Response Variable, Output Variable, Target Variable

A dependent variable is the quantity a model or experiment aims to explain or predict. Its value is treated as a function of one or more independent variables (features). In supervised learning, it is the labeled target: the class, score, next token, or continuous response.

Why it matters

  • Defines what the loss function compares predictions against
  • Choice of target (raw label vs transformed) shapes difficulty and metrics
  • Leakage into features can make a “dependent” variable look trivially predictable

How it shows up

In classical stats you often write $y = f(X) + \epsilon$. In ML pipelines the same idea is label / target columns. For language models, the dependent variable at each step is typically the next token (or a masked token in bidirectional models).

Keep naming precise in write-ups: “dependent variable,” “response,” and “target” are interchangeable; “output” can confuse people with model logits versus ground truth.

← Back to AI Glossary