Commit graph

8 commits

Author SHA1 Message Date
a10592847d feat(site_finder): future-supply-pressure index (#950 Step 6) (#1006)
All checks were successful
Deploy / changes (push) Successful in 5s
Deploy / build-frontend (push) Has been skipped
Deploy / build-backend (push) Successful in 1m20s
Deploy / build-worker (push) Successful in 2m23s
Deploy / deploy (push) Successful in 1m5s
2026-06-03 05:10:08 +00:00
900802264a feat(site-finder): supply_layers v2 compute service (#950 EPIC6 step3+4) (#1004)
All checks were successful
Deploy / build-worker (push) Successful in 2m30s
Deploy / changes (push) Successful in 5s
Deploy / build-frontend (push) Has been skipped
Deploy / build-backend (push) Successful in 1m40s
Deploy / deploy (push) Successful in 1m6s
2026-06-02 20:34:26 +00:00
59be55f80e feat(site-finder): per-competitor relevance_weight (#949 PR B) (#1000)
All checks were successful
Deploy / changes (push) Successful in 6s
Deploy / build-worker (push) Successful in 2m34s
Deploy / build-frontend (push) Has been skipped
Deploy / build-backend (push) Successful in 1m26s
Deploy / deploy (push) Successful in 1m10s
2026-06-02 19:45:38 +00:00
45d61ecff0 feat(site_finder): market-metrics service (#949 PR A, §9.2) (#997)
All checks were successful
Deploy / build-frontend (push) Has been skipped
Deploy / build-backend (push) Successful in 1m30s
Deploy / changes (push) Successful in 5s
Deploy / build-worker (push) Successful in 3m49s
Deploy / deploy (push) Successful in 1m10s
2026-06-02 19:26:26 +00:00
dbae4b0bda feat(site-finder): macro_indicator table + backfill + reader (#945 PR A) (#963)
All checks were successful
Deploy / build-frontend (push) Has been skipped
Deploy / build-worker (push) Successful in 2m20s
Deploy / build-backend (push) Successful in 2m32s
Deploy / deploy (push) Successful in 1m14s
Deploy / changes (push) Successful in 6s
2026-06-02 18:41:28 +00:00
lekss361
689327909c fix(sf-09): capacity-aware redistribute (round 2) — bounded + correct invariant
Round 1 (commit bcd7dc8) был broken: на 2-bucket входах surplus уходил в free
полностью без учёта capacity → free превышал cap → следующая итерация
clamp'ировала его и наоборот. Infinite oscillation в FastAPI handler.

Round 2 fix per review BLOCK (#282 comment):
- Surplus распределяется пропорционально available capacity (cap - v),
  не текущему v. Free никогда не вылетит выше cap.
- free = строго < cap (не <=) — иначе деление на 0 capacity.
- Hard guard `for _ in range(N+1)` — гарантированно завершается.
- Pathological (surplus > total_capacity): возвращаем оригинальный pct_map
  + cap_skipped=True (sum=100 invariant сохранён).
- Hamilton round вынесен в _hamilton_round() helper.

Tests:
- 2-bucket cases (90/10, 70/30, 99/1) expected cap_skipped=True
- test_cap_iteration_count_bounded — все pathological завершаются < 100ms
- All 13 cases verified standalone (3 fast-path + 7 reproduced + 3 pathological)
2026-05-17 15:30:50 +03:00
lekss361
bcd7dc8f75 fix(sf-09): iterative cap algorithm + cap_skipped flag + unit tests
Review verdict MINOR (PR #282) — single-pass cap_and_redistribute
не гарантировал invariant max ≤ MAX_BUCKET_SHARE_PCT. Surplus от clamped
поднимал free выше cap (например {1k:75, studio:15, 2k:10} → studio=39).

- backend/app/services/site_finder/best_layouts.py
  - _cap_and_redistribute → iterative while-loop до сходимости
  - Returns (result_map, cap_skipped) — флаг для pathological all-clamped
  - Float work map, Hamilton финальный pass для invariant sum=100
- backend/app/schemas/parcel.py
  - LayoutTzRecommendation.cap_skipped: bool = False
- backend/tests/services/site_finder/test_best_layouts.py
  - 6 parametrized tests: invariants, 7 failing cases from review,
    no-dominant unchanged, empty, cap_skipped propagation, normal-case
- frontend/src/types/best-layouts.ts — cap_skipped: boolean
- frontend/src/components/site-finder/BestLayoutsBlock.tsx
  - banner условие сменилось с maxPct > 60 на rec.cap_skipped (honest signal)
2026-05-17 15:21:51 +03:00
lekss361
5a4408e0ab fix(sf-01): time_window honest velocity — inline SQL с реальным фильтром report_month
Раньше _VELOCITY_DIVISORS делил агрегаты mv_layout_velocity (24 мес)
на 4/12 для quarter/year, не меняя реальное окно данных. Теперь
inline SQL из objective_corpus_room_month с CAST(:window_interval AS interval).

velocity_per_month = deals_window / months_in_window (1.0/3.0/12.0).
Разные time_window → разные строки из БД → разный mix/velocity/jk_count.

Closes (epic part) #271 item 1
2026-05-17 12:12:19 +03:00