mirror of
https://github.com/Hestia-Homes/Model.git
synced 2026-06-08 11:17:27 +00:00
docs: handover — flagged numbers were stale (different branch), Part 1 is the task
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
parent
8bd8ff8e5c
commit
1085842395
1 changed files with 11 additions and 38 deletions
|
|
@ -43,47 +43,20 @@ at 1e-4. The user will provide the cert numbers + drop the PDFs. For each:
|
|||
worksheet but cascade 0/wrong → calculator; input field absent in `epc` → mapper or
|
||||
extractor. Fix one cause = one slice.
|
||||
|
||||
### Part 2 — six flagged certs to fetch fresh and debug
|
||||
### Part 2 — (secondary) re-check the previously-flagged certs on THIS branch
|
||||
|
||||
The user's dashboard flags these (their numbers, **sign = lodged − our**):
|
||||
A dashboard once flagged six certs (0240, 0390-2954-3640, 2130, 6035, 7536, 9390). **Those
|
||||
numbers are STALE — they came from a branch WITHOUT this branch's fixes** (the user confirmed
|
||||
this). Do not chase them. On THIS branch the picture is different and mostly settled:
|
||||
|
||||
| cert | lodged | their "our" | their Δ |
|
||||
|---|---|---|---|
|
||||
| 0240-0200-5706-2365-8010 | 73 | 71.73 | +1.27 |
|
||||
| 0390-2954-3640-2196-4175 | 60 | 66.85 | −6.85 |
|
||||
| 2130-1033-4050-5007-8395 | 82 | 83.35 | −1.35 |
|
||||
| 6035-7729-2309-0879-2296 | 70 | 67.81 | +2.19 |
|
||||
| 7536-3827-0600-0600-0276 | 68 | 69.07 | −1.07 |
|
||||
| 9390-2722-3520-2105-8715 | 67 | 71.24 | −4.24 |
|
||||
- 7536 (68.924, +1), 2130 (83.78, +2), 0240 (−1) — concluded **0240-like** (API-only
|
||||
residuals; see per-cert notes below). 0390-2954-3640 pins at **+0** (exact).
|
||||
- 6035 (+2.19) and 9390 (community, −2) carry documented open residuals (see notes) but are
|
||||
lower-priority and not worksheet-backed.
|
||||
|
||||
### ⚠ CRITICAL — reconcile the numbers FIRST, before debugging
|
||||
|
||||
**The user's flagged numbers DO NOT match the golden-fixtures cascade.** All six certs are
|
||||
already golden fixtures (`tests/domain/sap10_calculator/rdsap/fixtures/golden/<cert>.json`),
|
||||
and the cascade gives different values:
|
||||
|
||||
- **0390-2954-3640 is pinned at resid +0** (our cascade = 60, EXACTLY lodged) — but the user
|
||||
flags it at **66.85 (−6.85)**. A 6.85 SAP gap can't be staleness.
|
||||
- 7536 (their 69.07) and 2130 (their 83.35) are **pre-this-session** values — the S0380.214
|
||||
roof fix moved 7536 → 68.924, and the S0380.215 wall fix moved 2130 → 83.78.
|
||||
|
||||
So the user's numbers come from a **different computation** than the golden cascade. Two
|
||||
hypotheses, test both before assuming the cascade is wrong:
|
||||
|
||||
1. **Fresh API JSON ≠ curated fixture.** The golden fixtures were bulk-fetched once
|
||||
(`scripts/fetch_cohort2_api_jsons.py`, which *skips certs whose JSON already exists`) and
|
||||
some may have been hand-corrected since. **Fetch each cert fresh and `diff` the raw JSON
|
||||
against the committed fixture.** If they differ, the fixture was curated and the fresh raw
|
||||
data is what the user's pipeline sees — debug the FRESH data. This is the most likely cause
|
||||
and exactly why the user wants a fresh fetch.
|
||||
2. **A different SAP engine.** The production stack (`backend/SearchEpc.py` →
|
||||
`etl/epc_clean/epc_attributes/*` → `backend/engine/engine.py`) is a SEPARATE mapping +
|
||||
scorer from `cert_to_inputs`. If the user's dashboard is produced there, that's a different
|
||||
code path than the golden cascade. Ask the user which pipeline the table came from.
|
||||
|
||||
Do NOT start "fixing" the cascade to hit the user's numbers until you know which pipeline
|
||||
produced them. The golden cascade is worksheet-validated for 47 certs; chasing a dashboard
|
||||
number from a different stack would regress it.
|
||||
So Part 2 is only worth touching if a **fresh fetch differs from the committed fixture**
|
||||
(curated/hand-corrected fixtures can mask raw-API mapper behaviour) — `diff` fresh vs fixture
|
||||
and debug the delta. Otherwise these are done; the real new work is **Part 1**.
|
||||
|
||||
---
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue