Merge pull request 'Бэктест оценщика берёт тип дома из справочника домов, как боевая оценка — штраф за тип дома впервые участвует в замере' (#3571) from fix/backtest-house-type into main

tradein-mvp/backend/scripts/backtest_estimator.py

@@ -1674,12 +1674,16 @@ class _HouseIdResolution:
     total: int = 0  # deals processed in the full-engine path under the flag
     resolved: int = 0  # deals where match_house_readonly returned an id
     imv_reachable: int = 0  # of resolved, those whose house_imv anchor fired
+    # #2862: сделкам, у которых тип дома подставлен из `houses` (deals.house_type пуст
+    # у всех строк). Ноль здесь = штраф за тип дома в прогоне не участвовал вовсе.
+    house_type_from_houses: int = 0
 
     def as_json(self) -> dict[str, int]:
         return {
             "resolved": self.resolved,
             "total": self.total,
             "imv_reachable": self.imv_reachable,
+            "house_type_from_houses": self.house_type_from_houses,
         }
 
 
@@ -2550,6 +2554,18 @@ def run_backtest_full(
             target_house_id = _resolve_house_id_for_deal(db, deal)
             if target_house_id is not None:
                 resolution.resolved += 1
+            # #2862: боевой estimate_quality (#3234) берёт тип дома из `houses`, когда
+            # его нет в форме — по резолвленному house_id, иначе ближайший дом в 60 м.
+            # deals.house_type пуст у всех сделок, так что без этого штраф за тип дома
+            # в прогоне не срабатывал ни разу. Только тип: год и этажность из `houses`
+            # сдвинули бы метрики по другим признакам и смешали бы замер.
+            if deal.house_type is None:
+                facts = est.m._lookup_house_facts(
+                    db, target_house_id=target_house_id, lat=deal.lat, lon=deal.lon
+                )
+                if facts is not None and facts.house_type is not None:
+                    deal = dataclasses.replace(deal, house_type=facts.house_type)
+                    resolution.house_type_from_houses += 1
         try:
             pr = _predict_full_spine(
                 db,
@@ -2616,10 +2632,11 @@ def run_backtest_full(
     if resolution is not None:
         metrics["house_id_resolution"] = resolution.as_json()
         logger.info(
-            "house_id resolution: resolved=%d/%d imv_reachable=%d",
+            "house_id resolution: resolved=%d/%d imv_reachable=%d house_type_from_houses=%d",
             resolution.resolved,
             resolution.total,
             resolution.imv_reachable,
+            resolution.house_type_from_houses,
         )
 
     if dump_fixture is not None and capture is not None:

tradein-mvp/backend/tests/test_backtest_estimator.py

@@ -519,13 +519,23 @@ def test_argparse_resolve_house_id_override() -> None:
 
 
 def test_house_id_resolution_as_json_shape() -> None:
-    res = bt._HouseIdResolution(total=10, resolved=4, imv_reachable=2)
-    assert res.as_json() == {"resolved": 4, "total": 10, "imv_reachable": 2}
+    res = bt._HouseIdResolution(total=10, resolved=4, imv_reachable=2, house_type_from_houses=3)
+    assert res.as_json() == {
+        "resolved": 4,
+        "total": 10,
+        "imv_reachable": 2,
+        "house_type_from_houses": 3,
+    }
 
 
 def test_house_id_resolution_defaults_zero() -> None:
     res = bt._HouseIdResolution()
-    assert res.as_json() == {"resolved": 0, "total": 0, "imv_reachable": 0}
+    assert res.as_json() == {
+        "resolved": 0,
+        "total": 0,
+        "imv_reachable": 0,
+        "house_type_from_houses": 0,
+    }
 
 
 # --------------------------------------------------------------------------- #
@@ -1006,3 +1016,108 @@ def test_predict_full_spine_passes_resolved_city_to_corridor(
     assert captured.get("called") is True
     assert captured.get("city") == "нижний тагил"
     assert pred is not None
+
+
+# --------------------------------------------------------------------------- #
+# #2862: тип дома из `houses` доезжает до отбора аналогов, как в estimate_quality.
+# --------------------------------------------------------------------------- #
+
+
+def _run_full_with_house_facts(
+    monkeypatch: pytest.MonkeyPatch, *, resolve_house_id: bool
+) -> tuple[dict[int, set[object]], list[dict[str, object]], dict[str, object]]:
+    from types import SimpleNamespace
+
+    from app.services import estimator as est_mod
+
+    def _deal(deal_id: int, house_type: str | None) -> object:
+        return bt.DealSample(
+            id=deal_id,
+            lon=60.6,
+            lat=56.8,
+            rooms=2,
+            sold_ppm2=100_000.0,
+            deal_date=None,
+            area_m2=50.0,
+            address="Екатеринбург, ул. Ленина, 5",
+            floor=3,
+            year_built=None,
+            house_type=house_type,
+        )
+
+    # Сделка 1 — как все сделки на проде (house_type пуст); сделка 2 — тип уже есть.
+    monkeypatch.setattr(bt, "_load_sample", lambda *a, **kw: [_deal(1, None), _deal(2, "brick")])
+    monkeypatch.setattr(bt, "_resolve_house_id_for_deal", lambda _db, _deal: 77)
+
+    lookups: list[dict[str, object]] = []
+
+    def _fake_lookup(_db: object, **kw: object) -> object:
+        lookups.append(kw)
+        return est_mod._HouseFacts(
+            house_id=77, total_floors=16, year_built=1985, house_type="panel", material_walls=None
+        )
+
+    seen: dict[int, set[object]] = {}
+    current: dict[str, int] = {}
+
+    def _fake_fetch_analogs(_db: object, **kw: object) -> tuple[list[object], bool, str]:
+        seen.setdefault(current["id"], set()).add(kw["house_type"])
+        return [], False, "W"
+
+    def _fake_price_from_inputs(**_kw: object) -> SimpleNamespace:
+        return SimpleNamespace(
+            median_price=5_000_000.0,
+            median_ppm2=100_000.0,
+            confidence="low",
+            anchor_tier=None,
+            expected_sold_per_m2=95_000.0,
+            expected_sold_price=4_750_000.0,
+            expected_sold_range_low=4_000_000.0,
+            expected_sold_range_high=5_500_000.0,
+        )
+
+    real_predict = bt._predict_full_spine
+
+    def _tracking_predict(db: object, deal: object, est: object, **kw: object) -> object:
+        current["id"] = deal.id  # type: ignore[attr-defined]
+        return real_predict(db, deal, est, **kw)  # type: ignore[arg-type]
+
+    monkeypatch.setattr(est_mod, "_lookup_house_facts", _fake_lookup)
+    monkeypatch.setattr(est_mod, "_fetch_analogs", _fake_fetch_analogs)
+    monkeypatch.setattr(est_mod, "_resolve_target_city", lambda _addr: None)
+    monkeypatch.setattr(est_mod, "_fetch_dkp_corridor", lambda *a, **kw: None)
+    monkeypatch.setattr(est_mod, "_fetch_anchor_comps", lambda *a, **kw: ([], None))
+    monkeypatch.setattr(est_mod, "_fetch_house_imv_anchor", lambda *a, **kw: None)
+    monkeypatch.setattr(est_mod, "_price_from_inputs", _fake_price_from_inputs)
+    monkeypatch.setattr(bt, "_predict_full_spine", _tracking_predict)
+
+    metrics = bt.run_backtest_full(
+        None,  # type: ignore[arg-type]
+        sample=2,
+        since="2025-06-01",
+        resolve_house_id=resolve_house_id,
+    )
+    return seen, lookups, metrics
+
+
+def test_full_backtest_takes_house_type_from_houses_like_prod(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    seen, lookups, metrics = _run_full_with_house_facts(monkeypatch, resolve_house_id=True)
+
+    # Пустой тип сделки заменён типом дома — именно он уходит в каждый отбор аналогов.
+    assert seen[1] == {"panel"}
+    # Свой тип сделки главнее справочника, как форма главнее houses на проде.
+    assert seen[2] == {"brick"}
+    assert lookups == [{"target_house_id": 77, "lat": 56.8, "lon": 60.6}]
+    assert metrics["house_id_resolution"]["house_type_from_houses"] == 1  # type: ignore[index]
+
+
+def test_full_backtest_without_flag_keeps_house_type_untouched(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    seen, lookups, metrics = _run_full_with_house_facts(monkeypatch, resolve_house_id=False)
+
+    assert seen[1] == {None}
+    assert lookups == []
+    assert "house_id_resolution" not in metrics