Latest summary
Reviewed LV fixed national candidate bundle at /home/synthestat/output/runs/LV/lv_population_review_national_candidate_2025_csb_freeze_44d91be2_seed420987 and returned PASS for the declared national/private-household candidate scope. Independent audits verified exact 1,860,565 person rows, 843,907 household/dwelling rows, zero age-sex residuals across 202 LV_CSP_IRD041 cells, zero minor-only/under-15-without-adult private households, zero under-25 reference persons, and zero YCH child-family households lacking an under-25 child member; unavailable building/hidden/work-school layers are explicit rather than fabricated.
Body
Country: LV — Latvia
Modeler fix from parent t_ed1b43de is ready for independent internal review.
Fixed/rebuilt bundle:
`/home/synthestat/output/runs/LV/lv_population_review_national_candidate_2025_csb_freeze_44d91be2_seed420987`
What changed:
- Repaired household assignment root cause: prior age-sorted stream assigned children/infants as P1 reference persons.
- New allocator reserves adult reference persons for all private households, assigns minors only into households with adults, downgrades impossible child-family type/size shells, and ensures retained child-family households include a plausible child member under 25.
- Added regression test: `/home/synthestat/tests/test_lv_national_household_model.py`.
- Updated builder: `/home/synthestat/workspace/manager_handoffs/modeller/build_lv_national_population_review_bundle.py`.
Modeler-reported verification:
- persons = 1,860,565; households = 843,907; dwellings = 843,907.
- LV_CSP_IRD041 age_source_code x sex max_abs residual = 0 across 202 cells.
- reference persons under 15/18/25 = 0 / 0 / 0.
- minor-only private households = 0.
- under-15/no-adult households = 0.
- child-family households = 181,494; child-family households without child under 25 = 0.
- private-household population relaxation remains explicit: +26,783 / 1.4605% against LV_CSP_MVS041.
Tests run:
- `python -m pytest tests/test_lv_national_household_model.py tests/test_synthesis_engine.py -q` -> 12 passed.
- `python -m pytest -q` -> 337 passed / 1 failed, known unrelated FR `degraded_zone_count` expectation.
Review acceptance focus:
- Independently audit no minor-only ordinary private households.
- Independently audit no underage reference persons.
- Independently audit child-family type/member semantics.
- Confirm exact HARD row count and age-sex controls remain intact.
- Confirm unavailable building/work/school/hidden overlays remain explicit and not fabricated.