What does embargoing do that purging does not?
- AIt removes rows whose label window crosses the split
- BIt drops a gap of training rows immediately after the validation set
- CIt shuffles the folds so each contains several regimes
- DIt reweights the training rows by how recent they are
Show the answer and worked solution
Answer: B – It drops a gap of training rows immediately after the validation set
Purging removes training rows whose label window overlaps the validation period, which handles the mechanical overlap. Embargoing goes further and drops a gap of training rows immediately after the validation set, because serial correlation means those rows still carry information about it even when their label windows no longer touch. In practice the embargo is set at a small percentage of the sample, tuned to how persistent the features are. Neither has anything to do with shuffling or with reweighting by recency.
- A. That is purging.
- B. Correct – serial correlation means those rows still carry information about the validation period.
- C. Shuffling is the thing being avoided.
- D. A reasonable technique, and unrelated to leakage.
Takeaway: Purge across the boundary; embargo after it.
Answer it in practice – your answer is marked and recorded.
Learn the method
Reported in interviews at
More machine learning questions
- A classifier has 30 true positives, 20 false positives and 10 false negatives.Foundation
- A classifier predicts probability 0.8 for an example that is in fact positive.Foundation
- Cross-validating a return forecaster, part 1 of 3Foundation
- Why is ordinary shuffled k-fold cross-validation wrong on a financial time series?Applied
- A classifier assigns scores at random. What is its expected AUC?Applied
- A classifier flags 500 cases, of which 40 are true positives, and misses 60 positives.Applied