Effect method changelog

The current method is wfm-effect-1.2. A published figure is never edited in place: when a change alters a number, the old row is superseded and the new one carries the new method id. Each entry says which rows were affected. The method itself is at /effect-method/.

Versions

  1. Clarification: a continuous series whose baseline median is zero has a zero noise floor. The engine used to divide by zero for it; it now reports insufficient_baseline. No published figure changes, because no figure could be computed for such a case.
  2. wfm-effect-1.2: national holidays (section 4): Chuseok (KR), Golden Week and Obon (JP), Songkran (TH) now exclude baseline days. Only rows whose baseline skips one of these days change, and they are recomputed and re-identified as 1.2 (the old row is superseded, not overwritten); all others are numerically unchanged. Prompted by a Pyeongtaek port-calls drop on 2026-09-23 to 2026-09-25 that the detector flagged and that was Chuseok.
  3. wfm-effect-1.1: ambient events (section 3). An earlier event in force for the whole baseline no longer excludes its days. Only rows that use this rule are re-identified as 1.1; all others are numerically identical to 1.0 and were not recomputed.
  4. wfm-effect-1.0: first version.

Placebo test. Corrections log.