data/sql/160_refresh_ekb_districts_median_dedup.sql

@@ -1,91 +0,0 @@
--- Fix: dedup deals-to-district assignment in ekb_districts median refresh.
---
--- Context: 67_refresh_ekb_districts_median.sql introduced ST_Intersects(district.geom, quarter.geom)
--- to join cadastral quarters to districts. A quarter whose polygon straddles a district boundary
--- matched BOTH districts, causing its deals to be counted in multiple districts and skewing
--- the PERCENTILE_CONT / AVG aggregates.
---
--- Fix approach: replace ST_Intersects with ST_Within(ST_Centroid(cq.geom), d.geom).
--- The centroid of each cadastral quarter falls in exactly one district polygon, guaranteeing
--- a 1:1 quarter→district assignment. A DISTINCT ON fallback (ordered by d.district_name) is
--- added in case district polygons overlap at all (they should not, but defensive).
---
--- Measured impact (prod, 2026-06-17, 24-month window):
---   ST_Intersects  → 2945 deal-rows (2482 quarter-district pairs, 85 duplicate quarter pairs)
---   ST_Within(centroid) → 2638 deal-rows (2396 quarter-district pairs, 0 duplicates)
---   Eliminated: 307 spurious deal-district assignments.
---
--- Dependencies: ekb_districts_geom (geom), cad_quarters_geom (geom, cad_number),
---               rosreestr_deals (quarter_cad_number, region_code, doc_type,
---               realestate_type_code, area, price_per_sqm, period_start_date),
---               ekb_districts (district_name, median_price_per_m2, mean_price_per_m2).
---
--- 67_refresh_ekb_districts_median.sql ran on prod; this file supersedes it via _schema_migrations.
--- No dependent views or FK constraints on the function itself.
--- Signature unchanged — all existing callers (analytics_refresh.py, beat_schedule.py, etc.)
--- continue to work without modification.
---
--- Deploy order: this file only (no backend code change required).
--- Idempotent: CREATE OR REPLACE FUNCTION — safe to re-run.
-
-BEGIN;
-
-CREATE OR REPLACE FUNCTION refresh_ekb_districts_median(window_months int DEFAULT 24, min_deals int DEFAULT 50)
-RETURNS TABLE(district_name text, deals_used bigint, median_pm numeric, mean_pm numeric)
-LANGUAGE plpgsql AS $$
-BEGIN
-    RETURN QUERY
-    WITH
-    -- Assign each cadastral quarter to exactly ONE district via centroid containment.
-    -- ST_Within(centroid, district_geom) is 1:1 whereas ST_Intersects matched border-
-    -- straddling quarters to multiple districts, inflating counts and skewing medians.
-    -- DISTINCT ON (cq.cad_number) with deterministic ORDER BY guards against the edge
-    -- case of overlapping district polygons.
-    quarter_district AS (
-        SELECT DISTINCT ON (cq.cad_number)
-               cq.cad_number,
-               d.district_name
-          FROM cad_quarters_geom cq
-          JOIN ekb_districts_geom d
-            ON ST_Within(ST_Centroid(cq.geom), d.geom)
-         WHERE d.district_name <> 'не определён'
-         ORDER BY cq.cad_number, d.district_name
-    ),
-    per_district AS (
-        SELECT qd.district_name,
-               COUNT(*)::bigint AS deals,
-               PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY rd.price_per_sqm)::numeric AS p_median,
-               AVG(rd.price_per_sqm)::numeric AS p_mean
-          FROM rosreestr_deals rd
-          JOIN quarter_district qd ON qd.cad_number = rd.quarter_cad_number
-         WHERE rd.region_code = 66
-           AND rd.doc_type = 'ДДУ'
-           AND rd.realestate_type_code = '002001003000'
-           AND rd.area BETWEEN 15 AND 200
-           AND rd.price_per_sqm BETWEEN 30000 AND 1000000
-           AND rd.period_start_date >= NOW() - (window_months || ' months')::interval
-         GROUP BY qd.district_name
-    ),
-    upd AS (
-        UPDATE ekb_districts e
-           SET median_price_per_m2 = ROUND(pd.p_median, 0),
-               mean_price_per_m2   = ROUND(pd.p_mean, 0)
-          FROM per_district pd
-         WHERE e.district_name = pd.district_name
-           AND e.district_name <> 'не определён'
-           AND pd.deals >= min_deals
-        RETURNING e.district_name, pd.deals, pd.p_median, pd.p_mean
-    )
-    SELECT u.district_name::text, u.deals, ROUND(u.p_median, 0), ROUND(u.p_mean, 0)
-      FROM upd u
-     ORDER BY u.district_name;
-END;
-$$;
-
-COMMENT ON FUNCTION refresh_ekb_districts_median(int, int) IS
-    'Refresh ekb_districts.median_price_per_m2 + mean_price_per_m2 из ДДУ-сделок rosreestr_deals. '
-    'Spatial join: ST_Within(ST_Centroid(cad_quarter), district) — каждый квартал ровно в одном районе '
-    '(fixes ST_Intersects double-count на граничных кварталах; issue #1352). '
-    'Окно по умолчанию 24 мес, минимум 50 сделок на район. Запускается Celery beat раз в месяц.';
-
-COMMIT;