fix(site-finder): раскрыть proxy-fraction в EKB-median velocity-бенчмарке (#2445-A4) (#2448)

backend/app/services/site_finder/velocity.py

@@ -95,6 +95,11 @@ class VelocityResult:
     proxy_used: bool = False
     # proxy_sqm_per_deal — допущение «м² на сделку» для fallback (45.0); None на objective/none.
     proxy_sqm_per_deal: float | None = None
+    # Audit #2445 A4: доля report_month-строк ЕКБ-бенчмарка (_get_ekb_median), где
+    # deals_total_vol_m2 IS NULL и объём фабриковался через count×45.0 (тот же proxy,
+    # что и rosreestr_fallback, но внутри city-wide нормализующей медианы). 0.0 —
+    # медиана целиком из реальных deals_total_vol_m2 (или бенчмарк не считался/fallback-константа).
+    ekb_median_proxy_fraction: float = 0.0
     # RANK 1 (DCF velocity schedule): абсорбция ОДНОГО типового конкурента района,
     # м²/мес = monthly_velocity_sqm / max(n_with_sales, 1). Это НЕ суммарная скорость
     # района (monthly_velocity_sqm), а темп выбытия одного участника рынка — им и
@@ -120,6 +125,7 @@ class VelocityResult:
             "objective_coverage_pct": self.objective_coverage_pct,
             "proxy_used": self.proxy_used,
             "proxy_sqm_per_deal": self.proxy_sqm_per_deal,
+            "ekb_median_proxy_fraction": round(self.ekb_median_proxy_fraction, 3),
             "project_absorption_sqm_per_month": (
                 round(self.project_absorption_sqm_per_month, 1)
                 if self.project_absorption_sqm_per_month is not None
@@ -307,9 +313,16 @@ def compute_velocity(
     # SF#1871 P2: surface objective-покрытие в payload (доля конкурентов с маппингом+данными).
     coverage_pct = round(mapped_ratio * 100, 1)
 
-    ekb_median = (
-        _get_ekb_median(db, months_window=months_window) or _EKB_MEDIAN_FALLBACK_SQM_PER_MONTH
-    )
+    # Audit #2445 A4: _get_ekb_median возвращает (median, proxy_row_fraction) — доля
+    # строк бенчмарка, где объём фабриковался через count*45.0. При DB-fallback
+    # (None — нет данных) proxy_fraction неизвестна → 0.0 (используем calibrated
+    # константу, а не contaminated медиану, так что disclosure не нужен).
+    _ekb_median_result = _get_ekb_median(db, months_window=months_window)
+    if _ekb_median_result is not None:
+        ekb_median, ekb_median_proxy_fraction = _ekb_median_result
+    else:
+        ekb_median = _EKB_MEDIAN_FALLBACK_SQM_PER_MONTH
+        ekb_median_proxy_fraction = 0.0
     n_comps = len(comp_rows)
     sample_no_data = sorted(
         [
@@ -342,6 +355,7 @@ def compute_velocity(
             ekb_median=ekb_median,
             sample_competitors=sample_no_data,
             objective_coverage_pct=coverage_pct,
+            ekb_median_proxy_fraction=ekb_median_proxy_fraction,
         )
         if rr_result is not None:
             return rr_result
@@ -363,6 +377,7 @@ def compute_velocity(
             objective_coverage_pct=coverage_pct,
             proxy_used=False,
             proxy_sqm_per_deal=None,
+            ekb_median_proxy_fraction=ekb_median_proxy_fraction,
             project_absorption_sqm_per_month=None,  # degenerate path — ill-defined
         )
 
@@ -470,6 +485,7 @@ def compute_velocity(
             ekb_median=ekb_median,
             sample_competitors=sample_no_data,
             objective_coverage_pct=coverage_pct,
+            ekb_median_proxy_fraction=ekb_median_proxy_fraction,
         )
         if rr_result is not None:
             return rr_result
@@ -503,6 +519,7 @@ def compute_velocity(
             objective_coverage_pct=coverage_pct,
             proxy_used=False,
             proxy_sqm_per_deal=None,
+            ekb_median_proxy_fraction=ekb_median_proxy_fraction,
             project_absorption_sqm_per_month=None,  # degenerate path — ill-defined
         )
 
@@ -580,6 +597,7 @@ def compute_velocity(
         objective_coverage_pct=coverage_pct,
         proxy_used=False,
         proxy_sqm_per_deal=None,
+        ekb_median_proxy_fraction=ekb_median_proxy_fraction,
         project_absorption_sqm_per_month=project_absorption,
     )
 
@@ -592,6 +610,7 @@ def _compute_rosreestr_fallback(
     ekb_median: float,
     sample_competitors: list[dict[str, Any]],
     objective_coverage_pct: float = 0.0,
+    ekb_median_proxy_fraction: float = 0.0,
 ) -> VelocityResult | None:
     """Fallback velocity через rosreestr_deals JOIN по cad_quarter участка.
 
@@ -681,13 +700,17 @@ def _compute_rosreestr_fallback(
         objective_coverage_pct=objective_coverage_pct,
         proxy_used=True,  # объём = deal_count × avg_area_per_deal (фабрикуется)
         proxy_sqm_per_deal=avg_area_per_deal,
+        # Audit #2445 A4: пробрасываем proxy-долю бенчмарка (ekb_median посчитан
+        # ДО вызова этого fallback) — даже здесь denominator может быть частично
+        # contaminated count*45.0-строками.
+        ekb_median_proxy_fraction=ekb_median_proxy_fraction,
         # RANK 1: квартальный total deal-count БЕЗ per-project декомпозиции —
         # per-project absorption ill-defined → None, НЕ драйвит DCF-график.
         project_absorption_sqm_per_month=None,
     )
 
 
-def _get_ekb_median(db: Session, months_window: int = 6) -> float | None:
+def _get_ekb_median(db: Session, months_window: int = 6) -> tuple[float, float] | None:
     """ЕКБ-wide медиана monthly velocity (м²/мес) per ЖК — benchmark.
 
     Источник: objective_corpus_room_month (актуальные данные, обновляется еженедельно).
@@ -697,6 +720,24 @@ def _get_ekb_median(db: Session, months_window: int = 6) -> float | None:
 
     Учитываются только ЖК с ≥3 месяцами данных за окно (стабильный сигнал).
     Fallback к _EKB_MEDIAN_FALLBACK_SQM_PER_MONTH если нет данных в БД.
+
+    Audit #2445 A4: report_month-строки с deals_total_count>0, но
+    deals_total_vol_m2 IS NULL, фабрикуют объём через COALESCE(..., count*45.0) —
+    тот же proxy-приём, что и в rosreestr_fallback (avg_area_per_deal), только
+    здесь он был необъявленным и контаминировал city-wide бенчмарк, к которому
+    нормализуется velocity_score КАЖДОГО участка. Строки не исключаются (иначе
+    продающийся, но без м²-данных ЖК просто выпадает из бенчмарка — тоже
+    искажение), но доля таких строк считается и возвращается вызывающему коду
+    для disclosure (см. VelocityResult.ekb_median_proxy_fraction), а не
+    проглатывается молча как чистые данные.
+
+    Возвращает (median, proxy_row_fraction) либо None если данных нет.
+    proxy_row_fraction — доля (project, room, month)-строк (среди вошедших в
+    медиану project'ов, ≥3 отчётных месяца), где объём взят из count*45.0,
+    а не из реального deals_total_vol_m2. Считается на уровне строк
+    objective_corpus_room_month (таблица хранит строку per project×room×month —
+    см. data/sql/100_fix_mv_layout_velocity_weighted_avg.sql), НЕ на уровне
+    report_month, чтобы не менять исходный HAVING-гейт «≥3 месяца данных».
     """
     try:
         with db.begin_nested():
@@ -704,23 +745,38 @@ def _get_ekb_median(db: Session, months_window: int = 6) -> float | None:
                 db.execute(
                     text(
                         """
-                    WITH per_project AS (
+                    WITH source_rows AS (
                         SELECT
                             project_name,
-                            SUM(COALESCE(deals_total_vol_m2,
-                                         deals_total_count * 45.0)) AS total_sqm,
-                            COUNT(DISTINCT report_month) AS months_data
+                            report_month,
+                            COALESCE(deals_total_vol_m2,
+                                     deals_total_count * 45.0) AS row_sqm,
+                            (deals_total_vol_m2 IS NULL)        AS is_proxy_row
                           FROM objective_corpus_room_month
                          WHERE report_month >= (CURRENT_DATE - CAST(:window_interval AS interval))
                            AND deals_total_count > 0
+                    ),
+                    per_project AS (
+                        SELECT
+                            project_name,
+                            SUM(row_sqm)                              AS total_sqm,
+                            COUNT(DISTINCT report_month)               AS months_data,
+                            COUNT(*)                                   AS rows_data,
+                            COUNT(*) FILTER (WHERE is_proxy_row)       AS proxy_rows
+                          FROM source_rows
                          GROUP BY project_name
                         HAVING COUNT(DISTINCT report_month) >= 3
                     ),
                     per_project_velocity AS (
-                        SELECT total_sqm / months_data AS velocity
+                        SELECT total_sqm / months_data AS velocity,
+                               rows_data,
+                               proxy_rows
                           FROM per_project
                     )
-                    SELECT PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY velocity) AS median
+                    SELECT
+                        PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY velocity) AS median,
+                        COALESCE(SUM(proxy_rows), 0) AS proxy_rows_total,
+                        COALESCE(SUM(rows_data), 0)  AS rows_total
                       FROM per_project_velocity
                     """
                     ),
@@ -734,6 +790,10 @@ def _get_ekb_median(db: Session, months_window: int = 6) -> float | None:
         # SAVEPOINT auto-rollback'нут
         return None
 
-    if row and row["median"] is not None:
-        return float(row["median"])
+    if row is None or row["median"] is None:
         return None
+
+    rows_total = int(row["rows_total"] or 0)
+    proxy_rows_total = int(row["proxy_rows_total"] or 0)
+    proxy_fraction = (proxy_rows_total / rows_total) if rows_total > 0 else 0.0
+    return float(row["median"]), proxy_fraction

backend/tests/test_velocity.py

@@ -126,7 +126,7 @@ def test_healthy_sales_returns_result():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -146,7 +146,7 @@ def test_few_competitors_low_confidence():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -164,7 +164,7 @@ def test_medium_confidence():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -184,6 +184,9 @@ def test_ekb_median_fallback_used_when_none():
 
     assert result is not None
     assert result.ekb_median_sqm == _EKB_MEDIAN_FALLBACK_SQM_PER_MONTH
+    # Audit #2445 A4: DB-fallback (no benchmark data at all) uses the calibrated
+    # constant, not a contaminated median -> proxy_fraction is 0.0 (nothing to disclose).
+    assert result.ekb_median_proxy_fraction == pytest.approx(0.0)
 
 
 def test_score_capped_at_1():
@@ -196,7 +199,7 @@ def test_score_capped_at_1():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -219,7 +222,7 @@ def test_score_zero_when_no_sales_sqm():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -247,6 +250,7 @@ def test_as_dict_structure():
         objective_coverage_pct=66.7,
         proxy_used=False,
         proxy_sqm_per_deal=None,
+        ekb_median_proxy_fraction=0.25,
     )
     d = vr.as_dict()
     assert "competitors_count" in d
@@ -265,6 +269,8 @@ def test_as_dict_structure():
     assert d["objective_coverage_pct"] == pytest.approx(66.7)
     assert d["proxy_used"] is False
     assert d["proxy_sqm_per_deal"] is None
+    # Audit #2445 A4: ekb_median_proxy_fraction disclosure field, rounded to 3dp.
+    assert d["ekb_median_proxy_fraction"] == pytest.approx(0.25)
     # RANK 1: project_absorption_sqm_per_month присутствует (None по умолчанию).
     assert "project_absorption_sqm_per_month" in d
     assert d["project_absorption_sqm_per_month"] is None
@@ -279,7 +285,7 @@ def test_sample_competitors_top5():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -306,7 +312,7 @@ def test_by_room_bucket_aggregation():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -330,7 +336,7 @@ def test_by_room_bucket_empty_when_no_bucket_data():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -356,7 +362,7 @@ def test_mapping_confidence_gate_in_sales_query():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -376,6 +382,66 @@ def test_fallback_constant_is_conservative():
     assert _EKB_MEDIAN_FALLBACK_SQM_PER_MONTH == pytest.approx(750.0)
 
 
+def test_ekb_median_proxy_fraction_surfaced_on_objective_path():
+    """Audit #2445 A4: когда _get_ekb_median сигнализирует, что часть строк
+    city-wide бенчмарка (COALESCE(deals_total_vol_m2, count*45.0)) фабриковалась,
+    эта доля должна долететь до VelocityResult.ekb_median_proxy_fraction —
+    а не проглатываться молча (проверяем и dataclass-поле, и as_dict()).
+    """
+    n = 6
+    comp_rows = [_comp_row(i) for i in range(1, n + 1)]
+    sales_rows = [_sales_row(i, total_sqm=4500.0, months=6) for i in range(1, n + 1)]
+    db = _make_db(comp_rows=comp_rows, sales_rows=sales_rows)
+
+    # 40% строк city-wide медианы — proxy (count*45.0), не реальный deals_total_vol_m2.
+    with patch(
+        "app.services.site_finder.velocity._get_ekb_median",
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.4),
+    ):
+        result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
+
+    assert result is not None
+    assert result.velocity_source == "objective"
+    assert result.ekb_median_proxy_fraction == pytest.approx(0.4)
+    assert result.as_dict()["ekb_median_proxy_fraction"] == pytest.approx(0.4)
+
+
+def test_ekb_median_proxy_fraction_surfaced_on_rosreestr_fallback():
+    """Audit #2445 A4: proxy_fraction бенчмарка пробрасывается и через
+    rosreestr_fallback путь (ekb_median посчитан ДО вызова fallback, contamination
+    в denominator остаётся релевантной вне зависимости от источника velocity)."""
+    comp_rows = [_comp_row(i) for i in range(1, 3)]
+    # Ниже _OBJECTIVE_COVERAGE_MIN_RATIO -> forces rosreestr_fallback path.
+    sales_rows = [_sales_row(1, total_sqm=0.0, months=0, has_mapping=False)]
+    db = _make_db(comp_rows=comp_rows, sales_rows=sales_rows)
+
+    rr_row = MagicMock()
+    start_d = datetime.date.fromisoformat("2025-01-01")
+    end_d = datetime.date.fromisoformat("2025-06-01")
+    rr_row.__getitem__ = lambda self, k: {
+        "total_deals": 60,
+        "period_start": start_d,
+        "period_end": end_d,
+    }[k]
+    rr_result_mock = MagicMock()
+    rr_result_mock.mappings.return_value.first.return_value = rr_row
+
+    # db.execute called: comp / sales / (rosreestr) — patch execute to append rr result.
+    original_side_effect = list(db.execute.side_effect)
+    db.execute.side_effect = [*original_side_effect, rr_result_mock]
+
+    with patch(
+        "app.services.site_finder.velocity._get_ekb_median",
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.6),
+    ):
+        result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT, cad_quarter="66:41:0702048")
+
+    assert result is not None
+    assert result.velocity_source == "rosreestr_fallback"
+    assert result.ekb_median_proxy_fraction == pytest.approx(0.6)
+    assert result.as_dict()["ekb_median_proxy_fraction"] == pytest.approx(0.6)
+
+
 def test_objective_path_surfaces_coverage_no_proxy():
     """SF#1871 P2 Fix2: objective-путь → proxy_used=False, coverage% посчитан, proxy_sqm=None."""
     n = 6
@@ -385,7 +451,7 @@ def test_objective_path_surfaces_coverage_no_proxy():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -449,7 +515,7 @@ def test_objective_path_sets_project_absorption_per_competitor():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -469,7 +535,7 @@ def test_objective_path_project_absorption_none_when_no_sales():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)
 
@@ -525,7 +591,7 @@ def test_sample_competitors_include_by_room_bucket():
 
     with patch(
         "app.services.site_finder.velocity._get_ekb_median",
-        return_value=_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH,
+        return_value=(_EKB_MEDIAN_FALLBACK_SQM_PER_MONTH, 0.0),
     ):
         result = compute_velocity(db, parcel_geom_wkt=_PARCEL_WKT)