A BCa interval is NaN for a metric that equals one on every resample. What should you investigate?
StatisticsLevel: IntermediateBootstrap uncertainty
A BCa interval is NaN for a metric that equals one on every resample. What should you investigate?
Hints
- Inspect the bootstrap distribution itself.
- A numerical warning can reveal a statistical limitation.
Answer
Inspect degeneracy, sample size and metric implementation. BCa cannot infer variability from a constant bootstrap distribution; another interval method does not create missing information.
Source and licence: SciPy — resampling and Monte Carlo inference · BSD-3-Clause · answer: Quantoros editorial