Use a cost-sensitive threshold: Which approach best addresses both the asymmetric error costs and limited

Asymmetric costs require cost-sensitive decisions; incomplete labels require targeted validation and explicit uncertainty rather than ordinary accuracy optimization.

The question

Preparing for AIGP? Take the free 5-min readiness quiz →

1. Train on all available transactions and select the model with highest precision.

   More transactions and precision do not resolve missing outcomes or establish performance for costly false negatives during demand surges.
2. Optimize overall accuracy using the existing labels.

   Overall accuracy can favor the common class and ignores both asymmetric stockout costs and uncertainty created by incomplete labels.
3. Use a cost-sensitive threshold, then validate recall and label uncertainty. ✓

   Cost-sensitive thresholding reflects the higher false-negative cost, while recall validation and uncertainty documentation address incomplete outcome evidence.
4. Increase the reorder threshold to reduce excess inventory exposure.

   A higher threshold may reduce over-ordering but increases missed surges, conflicting with the stated fivefold stockout cost.

How to remember it

Asymmetric costs require cost-sensitive decisions; incomplete labels require targeted validation and explicit uncertainty rather than ordinary accuracy optimization.

More Understanding How to Govern AI Development questions

• Measure each subgroup's false-positive rate against 5%: Which evaluation approach best satisfies that →
• Validate captions on representative data and test leakage: Which validation plan best addresses these →
• Test synthetic records for privacy leakage before use: Which action is most immediately necessary? →
• All 426 Understanding How to Govern AI Development questions →