What is Linear Discriminant Analysis (LDA)?

Updated May 16, 2026

Short answer

LDA is a supervised dimensionality reduction method that maximizes class separability.

Deep explanation

Unlike PCA, LDA uses class labels to maximize between-class variance and minimize within-class variance, making it useful for classification tasks.

Real-world example

Face recognition systems use LDA for class separation.

Common mistakes

  • Using LDA without labels.

Follow-up questions

  • How many components can LDA produce?
  • When is LDA better than PCA?

More Dimensionality Reduction interview questions

View all →