16 commits

Author	SHA1	Message	Date
bot-backend	d22e0e00ae	tech-debt(tradein/estimator): collapse won estimate_* flags into defaults (#1970) (#2475)	2026-07-12 12:46:18 +00:00
bot-backend	3505a4e3e5	feat(tradein/estimator): включить кросс-source физдедуп аналогов по умолчанию (#2087 H4, #2173) ... estimate_dedup_analogs_enabled: False → True. Бэктест #1966 OFF vs ON accuracy-идентичен (MAPE 13.89%, coverage 83.33%, bias −3.83%, median width/cv без изменений) — включение меняет только user-visible n_analogs: перестаёт быть раздутым кросс-постингом одного физлота на avito+cian+domklik ×3. Откат — ENV ESTIMATE_DEDUP_ANALOGS_ENABLED=false. Тесты разведены default vs explicit (не ослаблены): - test_estimator_dedup_cross_source_2087: добавлен test_dedup_default_is_on (без monkeypatch — дедуп активен по умолчанию); test_dedup_flag_off_is_noop оставляет ЯВНЫЙ OFF-override. - test_backtest_regression_gate: пиним флаг OFF — гейт это байт-идентичный replay frozen OFF-capture (recorded call-sequence фикстуры без дедупа); с ON _dedup_cross_source триммит listings до quarter_indexes_lookup и control-flow расходится с записью. - test_estimator_n_analogs_priced (autouse) + test_radius_path_n_analogs_unchanged: пиним OFF — инвариант «n_analogs = число priced-аналогов»/radius-passthrough ортогонален дедупу, а синтетические аналоги делят адрес/этаж/площадь и схлопнулись бы как кросс-посты. Refs #2087, #2173	2026-07-02 20:56:39 +03:00
lekss361	c45445b726	feat(tradein/estimator): hedonic year+area correction on expected_sold (#2002)	2026-06-27 16:06:08 +00:00
lekss361	9448a945d4	feat(tradein/estimator): calibrated 80% prediction interval for expected_sold range (#1966)	2026-06-27 14:01:24 +00:00
bot-backend	ec82742456	feat(tradein): same-building anchor safe-guard (#755) ... CLIENT-VISIBLE pricing (killer-factor, demo 2026-06-01). The same-building anchor replaced headline median/range/confidence with an over-confident premium on as few as 2 same-building asking comps. Human-chosen (Anton) 3-part safe-guard, all implemented: 1. min_comps 2 -> 4: anchor fires only with >=4 same-building comps. 2. confidence cap: at <5 anchor comps, confidence cannot be 'high' -> medium. 3. MAD-clip: drop price_per_m2 outliers among anchor comps before aggregation (k=estimate_sb_mad_k=3.5); if <min_comps survive, anchor does NOT fire (fallback to radius-median). Guards against price typos inflating anchor. Anchor still REPLACES the headline (uplift-no-replace variant NOT chosen). quarter-index/radius-tier/asking-sold untouched. 8 new unit tests; 61 anchor tests pass, ruff clean. MERGE-GATE (client-visible, like #764/#858): do NOT merge without a green backtest #763 (held-out MAPE <= 18.7% + canonical A/B Малышева 125 / Заводская 44а). Backtest needs live DB — pre-merge human/qa step. Refs #755, #694	2026-05-31 12:19:50 +03:00
bot-backend	34a13d71c3	fix(tradein): sync n_analogs to anchor count when anchor drives headline (#695) (#739) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 15:25:35 +00:00
bot-backend	2b3eb0b23d	fix(tradein): confidence_explanation no longer contradicts counters (#695) (#738) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 15:10:31 +00:00
bot-backend	a79cf7d5e0	fix(tradein): coarse-geo downgrade no longer shielded by Tier C anchor (#693) (#720) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 12:34:11 +00:00
bot-backend	0e2bba78b7	fix(tradein): coarse-geo downgrade works without DaData on prod (#693) (#718) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 12:08:56 +00:00
bot-backend	961a3caf2b	fix(tradein): surface same-building comps as analogs when anchor fires (#694) (#708) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 09:21:20 +00:00
bot-backend	9d94beae4f	fix(tradein): flag coarse-geo estimates as low-confidence (#693) (#707) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 08:58:23 +00:00
bot-backend	f04728297b	fix(tradein): run same-building anchor without radius analogs (#691) (#704) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 08:34:24 +00:00
bot-backend	6fac54642a	fix(tradein): distinct asking median vs expected-sold price (no double-discount) (#682) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 06:35:46 +00:00
bot-backend	3361e434d4	feat(tradein): refine same-building anchor for within-building heterogeneity (#681) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-30 05:49:46 +00:00
bot-backend	b5c26d2723	fix(tradein): harden same-building anchor address normalizer for prod formats (#677) (#679) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-29 22:21:29 +00:00
bot-backend	03a3296332	feat(tradein): same-building anchor estimator (#651/#652 v2) behind flag (#677) ... Co-authored-by: bot-backend <bot-backend@gendsgn.local> Co-committed-by: bot-backend <bot-backend@gendsgn.local>	2026-05-29 21:01:21 +00:00