What is train-test split in evaluation?

Updated May 17, 2026

Short answer

It divides data into training and testing sets to evaluate generalization.

Deep explanation

Training data is used to fit the model, while test data evaluates performance on unseen samples.

Real-world example

Used in image classification to ensure model works on new images.

Common mistakes

  • Leaking test data into training.

Follow-up questions

  • What is a good split ratio?
  • Why shuffle data?

More Model Evaluation interview questions

View all →