EasyDeepLearn

How do you pick the train/validation/test split sizes?

easy

Answer

  • Depends on total n and problem noise.
  • Small (n < 10k): 70/15/15 or use CV on train+val and hold out ~15% for final test.
  • Medium (10k-1M): 80/10/10.
  • Large (1M+): 98/1/1 is often enough — validation just needs a stable estimate of the metric (typically ≥ 10k samples).
  • Time series: contiguous, in temporal order, with test as the newest chunk.
  • Imbalanced: ensure enough minority-class examples in val and test — stratify.
Check yourself — multiple choice
  • Always 33/33/33
  • Scale by data size — small: 70/15/15; medium: 80/10/10; large: 98/1/1; stratify on imbalance
  • Split randomly regardless of order
  • Never split the data

Scale to data size; ensure enough minority-class samples in each split (stratify).

#cv#validation

Practise Supervised Learning

215 interview questions in this topic.

Related questions