Merge pull request 'fix(site): DfE LA averages for "vs LA avg"; no gap for independents (H2, part 2 of 2)' (#186) from fix/la-averages-site into main

Reviewed-on: #186

backend/app.py

@@ -1273,17 +1273,63 @@ async def get_filter_options(request: Request):
     }
 
 
+def _la_averages_payload(df: pd.DataFrame) -> dict:
+    """Per-LA Attainment 8 for the "vs LA avg" comparison: DfE's own LA
+    averages (fact_ks4_la_averages, all state-funded schools), never a mean of
+    the dataframe. That mean counted independent and special schools and put
+    most LAs about 7 points low (audit H2).
+
+    The year is the latest with any school Attainment 8, so the average and
+    the scores set against it are the same year. Figures are keyed by our LA
+    name through the LA code. An LA without a DfE figure for that year (City of
+    London) is absent; no figures for the year, or no mart, give an empty map,
+    so rows show no comparison, never another year's figure.
+    """
+    empty = {"year": 0, "secondary": {"attainment_8_by_la": {}}}
+    if df.empty or "attainment_8_score" not in df.columns:
+        return empty
+    scored = df[df["attainment_8_score"].notna()]
+    if scored.empty:
+        return empty
+    year = int(scored["year"].max())
+
+    la = (df[["local_authority_code", "local_authority"]]
+          .dropna()
+          .drop_duplicates("local_authority_code"))
+    name_by_code = {int(code): name for code, name in
+                    zip(la["local_authority_code"], la["local_authority"])}
+
+    from . import database
+    from .models import Ks4LaAverage
+
+    rows: list = []
+    db = None
+    try:
+        db = database.SessionLocal()
+        rows = db.query(Ks4LaAverage).filter(Ks4LaAverage.year == year).all()
+    except Exception:
+        import logging
+        logging.getLogger(__name__).warning(
+            "DfE LA averages unavailable for %s", year, exc_info=True)
+        if db is not None:
+            db.rollback()
+    finally:
+        if db is not None:
+            db.close()
+
+    by_la = {
+        name_by_code[row.la_code]: row.attainment_8_score
+        for row in rows
+        if row.attainment_8_score is not None and row.la_code in name_by_code
+    }
+    return {"year": year, "secondary": {"attainment_8_by_la": by_la}}
+
+
 @app.get("/api/la-averages")
 @limiter.limit(f"{settings.rate_limit_per_minute}/minute")
 async def get_la_averages(request: Request):
-    """Get per-LA average Attainment 8 score for secondary schools in the latest year."""
+    """DfE's per-LA Attainment 8 averages for the latest year with results."""
-    df = load_school_data()
+    return _la_averages_payload(load_school_data())
-    if df.empty:
-        return {"year": 0, "secondary": {"attainment_8_by_la": {}}}
-    latest_year = int(df["year"].max())
-    sec_df = df[(df["year"] == latest_year) & df["attainment_8_score"].notna()]
-    la_avg = sec_df.groupby("local_authority")["attainment_8_score"].mean().round(1).to_dict()
-    return {"year": latest_year, "secondary": {"attainment_8_by_la": la_avg}}
 
 
 _KS2_NATIONAL_METRICS = [

backend/models.py

@@ -344,6 +344,18 @@ class Ks4NationalAverage(Base):
     gcse_grade_91_pct = Column(Float)
 
 
+class Ks4LaAverage(Base):
+    """Official DfE KS4 local-authority averages (all state-funded schools) —
+    one row per academic year and LA. la_code is the GIAS LA code."""
+    __tablename__ = "fact_ks4_la_averages"
+    __table_args__ = MARTS
+
+    year = Column(Integer, primary_key=True)
+    la_code = Column(Integer, primary_key=True)
+    la_name = Column(String)
+    attainment_8_score = Column(Float)
+
+
 class Ks2NationalAverage(Base):
     """Official DfE KS2 national headline averages — one row per academic year."""
     __tablename__ = "fact_ks2_national_averages"

backend/tests/test_la_averages_marts.py

@@ -0,0 +1,119 @@
+"""/api/la-averages serves DfE's own LA averages (fact_ks4_la_averages).
+
+It used to average the dataframe: every school with an Attainment 8,
+independent and special schools included. Kensington and Chelsea came out at
+35.2 against DfE's 54.5, and most LAs about 7 points low (audit H2).
+"""
+
+import numpy as np
+import pandas as pd
+import pytest
+from fastapi.testclient import TestClient
+
+LATEST = 202425
+
+
+def _df():
+    return pd.DataFrame([
+        # A state school and an independent: their mean, 37.65, is not DfE's figure.
+        dict(year=LATEST, local_authority="Kensington and Chelsea", local_authority_code=207, attainment_8_score=54.9),
+        dict(year=LATEST, local_authority="Kensington and Chelsea", local_authority_code=207, attainment_8_score=20.4),
+        dict(year=LATEST, local_authority="Bristol, City of", local_authority_code=801, attainment_8_score=45.0),
+        dict(year=LATEST, local_authority="West Sussex", local_authority_code=938, attainment_8_score=48.0),
+        # DfE publishes no LA figure for City of London.
+        dict(year=LATEST, local_authority="City of London", local_authority_code=201, attainment_8_score=30.0),
+        # A newer year with primary results only.
+        dict(year=202526, local_authority="Kensington and Chelsea", local_authority_code=207, attainment_8_score=np.nan),
+    ])
+
+
+class _Row:
+    def __init__(self, year, la_code, la_name, attainment_8_score):
+        self.year = year
+        self.la_code = la_code
+        self.la_name = la_name
+        self.attainment_8_score = attainment_8_score
+
+
+class _StubSession:
+    rows = [
+        _Row(LATEST, 207, "Kensington and Chelsea", 54.5),
+        _Row(LATEST, 801, "Bristol City", 46.3),   # DfE's spelling, not ours
+        _Row(LATEST, 938, "West Sussex", None),    # suppressed
+        _Row(LATEST, 330, "Birmingham", 44.0),     # no school of ours there
+        _Row(202324, 207, "Kensington and Chelsea", 54.5),
+    ]
+
+    def query(self, model):
+        assert model.__name__ == "Ks4LaAverage"
+        return self
+
+    def filter(self, condition):
+        # The payload filters on year == <year>; apply it as Postgres would.
+        self._year = condition.right.value
+        return self
+
+    def all(self):
+        return [r for r in self.rows if r.year == self._year]
+
+    def rollback(self):
+        pass
+
+    def close(self):
+        pass
+
+
+class _OldYearOnly(_StubSession):
+    rows = [_Row(202324, 207, "Kensington and Chelsea", 54.5)]
+
+
+class _NoMart(_StubSession):
+    def all(self):
+        raise RuntimeError('relation "marts.fact_ks4_la_averages" does not exist')
+
+
+@pytest.fixture()
+def payload(monkeypatch):
+    from backend import app as app_module
+    from backend import database as database_module
+
+    def _run(session_cls):
+        monkeypatch.setattr(database_module, "SessionLocal", session_cls)
+        return app_module._la_averages_payload(_df())
+
+    return _run
+
+
+def test_serves_dfe_figures_keyed_by_our_la_names(payload):
+    out = payload(_StubSession)
+    assert out["year"] == LATEST
+    assert out["secondary"]["attainment_8_by_la"] == {
+        "Kensington and Chelsea": 54.5,
+        "Bristol, City of": 46.3,
+    }
+
+
+def test_an_la_without_a_dfe_figure_is_absent(payload):
+    by_la = payload(_StubSession)["secondary"]["attainment_8_by_la"]
+    assert "City of London" not in by_la
+    assert "West Sussex" not in by_la
+    assert None not in by_la.values()
+
+
+def test_no_dfe_figures_for_the_year_give_an_empty_map(payload):
+    assert payload(_OldYearOnly) == {"year": LATEST, "secondary": {"attainment_8_by_la": {}}}
+
+
+def test_a_missing_mart_gives_an_empty_map(payload):
+    assert payload(_NoMart)["secondary"]["attainment_8_by_la"] == {}
+
+
+def test_the_endpoint_serves_dfe_figures(monkeypatch):
+    from backend import app as app_module
+    from backend import database as database_module
+
+    monkeypatch.setattr(app_module, "load_school_data", _df)
+    monkeypatch.setattr(database_module, "SessionLocal", _StubSession)
+    resp = TestClient(app_module.app).get("/api/la-averages")
+    assert resp.status_code == 200, resp.text
+    assert resp.json()["secondary"]["attainment_8_by_la"]["Kensington and Chelsea"] == 54.5

docs/ARCHITECTURE.md

@@ -61,6 +61,12 @@ history from the full DataFrame and supplementary data from marts. Comparisons
 batch supplementary queries across selected URNs. Async routes still contain
 synchronous dependency calls; a fully asynchronous database layer is not present.
 
+DfE's official benchmarks live in marts: `fact_ks2_national_averages` and
+`fact_ks4_national_averages` for England, and `fact_ks4_la_averages` (all
+state-funded schools, per LA) for the search rows' "vs LA avg".
+`data_loader.compute_benchmarks` computes further state-school benchmarks from
+our own dataset; they are not DfE figures and are labelled as computed.
+
 ## Frontend boundaries
 
 `app/(frontend)` owns the public root layout and pages. `app/(payload)` owns the

docs/superpowers/specs/2026-10-06-ks4-2023-24-and-la-averages-design.md

@@ -0,0 +1,300 @@
+# 2023/24 Results and DfE LA Averages — Design
+
+**Date:** 2026-10-06
+**Status:** approved design, not yet implemented
+**Scope:** `tap_uk_ees` (KS4 results, KS4 information, new LA stream),
+`safe_numeric`, new `fact_ks4_la_averages` mart, annual EES DAG selector,
+`/api/la-averages`, secondary search rows and map cards
+**Fixes:** audit findings C2 and H2, from the 3 Oct 2026 accuracy audit
+
+## Goal
+
+Show every 2023/24 result and school-information figure DfE published, and
+compare each state-funded secondary school with DfE's own local-authority
+average.
+
+## The problem
+
+### C2: 2023/24 is empty
+
+Bishop Stopford School (137086) shows Attainment 8 60.7 for 2022/23, nothing
+for 2023/24 and 58.7 for 2024/25. DfE's 2023/24 figures are Attainment 8 64.1,
+Progress 8 +1.02 and English and maths grade 4+ 91.7%. 2023/24 is the last year
+DfE published Progress 8 (2024/25 has no KS2 baseline), so the site shows no
+recent Progress 8 for any school. Site-wide, 4,170 listed schools have a DfE
+2023/24 Attainment 8 and 3,384 a Progress 8.
+
+There are three separate causes.
+
+1. **KS4 results.** The 2024/25 release's
+   `202425_performance_tables_schools_final.csv` is a time series: it holds
+   2022/23, 2023/24 and 2024/25 under the current column names, with the right
+   values (Bishop Stopford 2023/24: 64.1, 1.02, 91.7). The 2023/24 release's own
+   `202324_performance_tables_schools_final.csv`, re-issued on 10 March 2026,
+   uses the older names (`t_pupils`, `avg_att8`, `avg_p8score`,
+   `pt_l2basics_94` …). `EESDatasetStream` reads releases in the API's order,
+   newest first, so the old file is read last. Its rows carry none of the
+   declared fields, and target-postgres upserts on the stream's primary key
+   (`append_only = not key_properties` in meltanolabs-target-postgres 0.8.0),
+   so they overwrite the good 2023/24 rows with nulls.
+
+   The two files share the keys of every "Total" row, but 34,254 of the 57,090
+   2023/24 sub-group rows use different labels ("Low prior" against "Low prior
+   attainment"). Those old-label rows sit in `raw.ees_ks4_performance` with
+   null measures. Nothing reads them.
+
+2. **KS4 school information.** 2023/24 information exists only in the 2023/24
+   release, in `202324_information_about_schools_final.csv`, with the older
+   names. `EESKS4InfoStream` declares the newer ones, so prior attainment,
+   SEN percentages, disadvantage gaps and Progress 8 banding are null for
+   2023/24.
+
+3. **KS2 school information.** The 2023/24 file
+   `ks2_school_information_data.csv` uses the declared names, and pupil counts
+   load (school 147411: 818 pupils, 112 eligible). Its percentages are written
+   with a sign (`ptfsm6cla1a = "34%"`). `safe_numeric` accepts only
+   `^-?[0-9]+(\.[0-9]+)?$`, so disadvantaged, EAL, SEN and mobility percentages
+   are null for every school in 2023/24.
+
+DfE published no school-level KS2 information file for 2022/23 (the 2023/24
+release's attainment file carries 2022/23 attainment rows only). Those nulls are
+a gap in the source, not a defect.
+
+### H2: "vs LA avg" uses the wrong average
+
+`/api/la-averages` (`backend/app.py`) takes an unweighted mean of every school
+in the LA with an Attainment 8 score, independent and special schools included.
+Independent schools score low because DfE measures exclude IGCSEs, and special
+schools score low for other reasons, so the average is too low almost
+everywhere. Kensington and Chelsea's "LA avg" is 35.2: the mean of 6 state
+schools (54.9) and 8 independent schools (20.4). DfE's figure is 54.5. Of the
+151 LAs the audit compared, ours was lower in 147, by 7.1 points on average and
+by up to 19.3, so most secondary schools look better than their area.
+
+DfE's LA averages are already in the file the pipeline downloads for the
+England averages: the "summary, all state-funded" data set
+(`data-catalogue/data-set/1b649e16-01e8-435b-a814-56be2faf9054/csv`). Its
+`Local authority` / `All state-funded` / Total rows match DfE's published
+performance-table LA averages (RECTYPE 4) for all 152 LAs in 2024/25, with no
+difference. `EESKs4NationalStream` keeps the England row and discards them.
+"All state-funded" is the same population as the England benchmark the site
+already shows.
+
+Computing the average ourselves from state-funded schools, weighted by pupils,
+was tested and rejected: it differs from DfE's figure by 1.1 points on average
+and is never exact.
+
+## Non-goals
+
+- 2022/23 KS2 school information (DfE published none).
+- An "excludes IGCSEs" note wherever an independent school's Attainment 8
+  appears. This change only stops comparing independent schools with the LA.
+- Showing 2023/24 Progress 8 in the GCSE section's headline. The history section
+  shows it once the data loads; the 2024/25 banner stays true.
+- Updating the fixed EES data-set id when DfE publishes 2025/26. It already
+  feeds the England averages; the LA stream shares it.
+- LA comparisons for other measures or on other pages.
+- Deleting the leftover old-label raw rows automatically.
+
+## The rules
+
+### The newest release owns every year it contains (KS4 results only)
+
+`EESDatasetStream` gets an opt-in class attribute,
+`_newest_release_owns_period: bool = False`. `EESKS4PerformanceStream` sets it
+to `True`. When it is on:
+
+- releases are processed newest first by `time_period` (from the release slug),
+  not in the API's order;
+- the stream records each `time_period` it has emitted;
+- in each older release, rows whose `time_period` a newer release already
+  emitted are dropped, and the stream logs how many it skipped and for which
+  years;
+- years only an older release contains are emitted as before.
+
+The filter is a pure function, testable without a download. Other streams keep
+the current behaviour. A general rule would be wrong: the 2024/25 KS2 file holds
+98,448 of the 955,956 rows the 2023/24 release has for 2023/24.
+
+### KS4 information: old names
+
+`EESKS4InfoStream._column_renames` maps the 2023/24 names onto the declared
+fields:
+
+| 2023/24 column | Declared field |
+|---|---|
+| `t_allks_pupils` | `allks_pupil_count` |
+| `t_allks_boys` | `allks_boys_count` |
+| `t_allks_girls` | `allks_girls_count` |
+| `t_pupils` | `endks4_pupil_count` |
+| `avg_ks2_scaledscore` | `ks2_scaledscore_average` |
+| `pt_sen_with_ehcp` | `sen_with_ehcp_pupil_percent` |
+| `pt_sen` | `sen_pupil_percent` |
+| `pt_sen_no_ehcp` | `sen_no_ehcp_pupil_percent` |
+| `diffn_att8` | `attainment8_diffn` |
+| `diffn_p8mea` | `progress8_diffn` |
+| `p8_banding` | `progress8_banding` |
+
+Newer files contain none of the old names, so they are unaffected.
+
+### `safe_numeric` accepts a trailing `%`
+
+The pattern becomes `^-?[0-9]+(\.[0-9]+)?%?$` and the cast reads
+`rtrim(col, '%')`. A value such as `34%` can only mean 34. Suppression codes
+(`c`, `z`, `x` …) still become null.
+
+### LA averages
+
+A new stream, `ees_ks4_la`, reads the same CSV as `ees_ks4_national` and keeps
+rows where `geographic_level = 'Local authority'`,
+`establishment_type_group = 'All state-funded'`, `breakdown_topic = 'Total'` and
+`breakdown = 'Total'` (case-insensitive, as the national stream compares). It
+emits `time_period`, `old_la_code`, `new_la_code`, `la_name` and the 8 headline
+measures the national stream emits (`_KS4_NATIONAL_COL_MAP`). Primary key:
+(`time_period`, `old_la_code`). The two streams share one download-and-filter
+helper.
+
+`old_la_code` is the GIAS LA code (`local_authority_code`), so schools join on
+the code, not the name. Names match today for every LA DfE publishes; DfE
+publishes no figure for City of London.
+
+### Which schools get a gap
+
+The search row and the map card show "vs LA avg" only when the school has an
+Attainment 8 score, is neither special (`isSpecialSchool`) nor independent
+(`isIndependentSchool`: "independent" in the GIAS type, which covers "Other
+independent school" and "Other independent special school"), and its LA has a
+DfE figure for the year the endpoint serves.
+
+## Delivery
+
+Two PRs, as for C1.
+
+### PR 1: pipeline
+
+- `pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py`: release
+  precedence, KS4 information renames, `ees_ks4_la` stream registered in
+  `discover_streams`, shared national/LA helper.
+- `pipeline/transform/macros/safe_numeric.sql`: trailing `%`.
+- `pipeline/transform/models/staging/`: source `raw.ees_ks4_la` and
+  `stg_ees_ks4_la` (view): `cast(old_la_code as integer) as la_code`,
+  `cast(time_period as integer) as year`, `la_name`, measures via
+  `safe_numeric`, named as in `stg_ees_ks4_national`.
+- `pipeline/transform/models/marts/fact_ks4_la_averages.sql` (table): one row
+  per (`year`, `la_code`) with `la_name` and the columns of
+  `fact_ks4_national_averages`. Schema tests: unique (`year`, `la_code`);
+  `year`, `la_code` and `la_name` not null.
+- Data tests in `pipeline/transform/tests/`:
+  - `assert_ks4_years_have_results`: every year in `stg_ees_ks4` has a non-null
+    Attainment 8 for at least 50% of its rows. DfE's files reach 82% each year;
+    2023/24 loads at 0% today. Pre-2019 years come from the legacy model and are
+    not tested.
+  - `assert_ks2_info_percentages_loaded`: for every year in `stg_ees_ks2` where
+    at least 1,000 rows have `total_pupils`, at least 90% of those rows have
+    `disadvantaged_pct`. DfE's files reach 96–97%. 2022/23 has no pupil counts
+    and is skipped.
+  - `assert_ks4_la_averages_cover_las`: the latest year in
+    `fact_ks4_la_averages` has at least 145 LAs (DfE: 152).
+- `pipeline/dags/school_data_pipeline.py`: the annual EES build selects
+  `stg_ees_ks4_la+`.
+- `docs/ARCHITECTURE.md`: LA averages come from DfE's data set.
+
+### PR 2: backend and UI (after the EES DAG has run on PR 1)
+
+- `backend/models.py`: `Ks4LaAverage` for `marts.fact_ks4_la_averages`.
+- `backend/app.py` `/api/la-averages`: the year is the latest with any school
+  Attainment 8 (as now). It reads that year's rows from the mart and keys each
+  `attainment_8_score` by our LA name, through the `local_authority_code` →
+  `local_authority` pairs in the school data. The response shape is unchanged.
+  No rows for that year, a missing table or a query error give an empty map,
+  logged, never another year's figures and never a computed mean.
+- `nextjs-app/lib/utils.ts`: `isIndependentSchool(school)`.
+- `nextjs-app/components/SecondarySchoolRow.tsx` and
+  `nextjs-app/components/LeafletMapInner.tsx`: the rule in "Which schools get a
+  gap".
+- `e2e/tests/journeys.spec.ts`: the two journeys under Testing.
+
+## Testing
+
+**Extractor (pytest, `pipeline/tests/`, new):**
+
+- precedence: with the flag on, a year in a newer release is emitted once, from
+  the newer release; a year only an older release has is kept; with the flag
+  off, every row passes;
+- releases arriving oldest first are still processed newest first;
+- KS4 information: an old-format row yields `endks4_pupil_count`,
+  `ks2_scaledscore_average`, `progress8_banding` and the rest of the table;
+- LA filter: a small CSV with national, regional, LA and sub-group rows yields
+  one row per LA and year, with the declared fields.
+
+**dbt (local `pgserver`, as for C1):**
+
+- unit test on `stg_ees_ks2`: `34%` → 34, `34` → 34, `c` → null;
+- unit test on `stg_ees_ks4_la` or the mart: codes and years cast, measures
+  carried;
+- the three data tests and the schema tests above;
+- `pipeline/tests/test_dag_selectors.py` passes with the new selector.
+
+**Backend (pytest):**
+
+- the response gives the mart's figure where the fixture's plain mean differs;
+- matching works by code when the mart's `la_name` differs from ours;
+- a mart without the served year, and a missing table, give an empty map.
+
+**Front end (Jest):**
+
+- `isIndependentSchool` for "Other independent school", "Other independent
+  special school", "Academy converter" and null;
+- `SecondarySchoolRow` and the map card show no gap for an independent school
+  and show one for a state school with an LA figure.
+
+**E2E (`journeys.spec.ts`, PR 2):**
+
+- C2: Bishop Stopford (137086) history shows 2023/24 Attainment 8 64.1 and
+  Progress 8 +1.02. These are final figures, so the journey stays stable.
+- H2: a search returning an independent and a state secondary: the independent
+  row has no "vs LA avg", the state row has one. The exact gap is not asserted:
+  DfE's 2025/26 provisional KS4 data is due and would change it.
+
+## Rollout and verification
+
+1. PR 1 merges to staging. Tudor runs `school_data_annual_ees` on staging.
+2. Through the staging API: Bishop Stopford 2023/24 Attainment 8 64.1 and
+   Progress 8 1.02; school 147411 2023/24 disadvantaged 34; the LA mart has 152
+   LAs for 2024/25. If a data test fails, the build stops before search sync;
+   investigate before going further.
+3. PR 2 merges, so the post-merge E2E gate runs against loaded data.
+4. Production, Tudor's decision, in order: promote PR 1, run the EES DAG on
+   production, promote PR 2. If PR 2 arrives first, the endpoint returns an
+   empty map and rows show no gap, never a wrong one.
+5. Optional cleanup, in the PR 1 description for Tudor:
+   `delete from raw.ees_ks4_performance where time_period = '202324' and pupil_count is null`.
+   New-format rows always have `pupil_count` (suppressed values are `c` or `z`,
+   not null), so this removes exactly the old-label leftovers.
+6. Validation on production against DfE's files: every listed school's 2023/24
+   Attainment 8, Progress 8 and English and maths figures match; the 2023/24 KS4
+   and KS2 information fields match; `/api/la-averages` equals DfE for all 152
+   LAs; independent rows show no gap. Then mark C2 and H2 resolved in the audit
+   report.
+
+## Expected visible change
+
+- Secondary history charts and tables run unbroken from 2022/23 to 2024/25, and
+  2023/24 Progress 8 appears for about 3,400 schools.
+- 2023/24 school information (KS2 and KS4) fills in.
+- "vs LA avg" drops for most state secondaries. In Kensington and Chelsea, a
+  school with Attainment 8 60.0 moves from +24.8 to +5.5.
+- Independent schools and City of London schools show no LA gap.
+
+## Risks
+
+- **Data-test thresholds stop a good build.** They were set from DfE's own
+  files (82% and 96–97% against 50% and 90%). A failure means the data changed
+  shape, which is what they are for.
+- **`safe_numeric` is shared by 14 models.** Values written as `n%` were null
+  and become numbers. No DfE column uses `%` for anything but a percentage.
+  The staging EES run is the check.
+- **Precedence drops data an older release holds more completely.** It is
+  opt-in for KS4 results, where both files hold the same 57,090 2023/24 keys.
+- **The fixed EES data-set id goes stale** when 2025/26 is published. The
+  endpoint's year check then gives no gap rather than a mismatched one.

e2e/tests/journeys.spec.ts

@@ -417,12 +417,15 @@ test('a secondary search row compares its Attainment 8 with the LA average', asy
   // guards the comparison itself; the unit test pins the cache mode.
   const la = await (await page.request.get('/api/la-averages')).json();
   const averages: Record<string, number> = la.secondary?.attainment_8_by_la ?? {};
+  // DfE publishes about 152 LA averages. An empty map (a missing mart, or a
+  // year the LA data set has not reached) hides every comparison: fail, not skip.
+  expect(Object.keys(averages).length).toBeGreaterThan(100);
   const res = await page.request.get('/api/schools?search=school&phase=secondary&page_size=50');
   expect(res.ok()).toBeTruthy();
   const school = ((await res.json()).schools ?? []).find(
     (s: { attainment_8_score?: number | null; local_authority?: string; school_type?: string }) =>
       s.attainment_8_score != null && s.local_authority != null && averages[s.local_authority] != null
-      && !/special|pupil referral|alternative provision/i.test(s.school_type ?? ''));
+      && !/special|pupil referral|alternative provision|independent/i.test(s.school_type ?? ''));
   test.skip(!school, 'no mainstream secondary with an LA average here');
 
   await searchByName(page, school.school_name);
@@ -433,6 +436,52 @@ test('a secondary search row compares its Attainment 8 with the LA average', asy
   await expect(stats.getByText(/vs LA avg/)).toBeVisible();
 });
 
+test('an independent secondary shows its Attainment 8 without an LA comparison', async ({ page }) => {
+  // DfE's LA averages cover state-funded schools, and an independent school's
+  // Attainment 8 leaves out IGCSEs, so a gap would mislead (audit H2). A state
+  // school on the same page must show its gap first, so the absence is real.
+  const la = await (await page.request.get('/api/la-averages')).json();
+  const averages: Record<string, number> = la.secondary?.attainment_8_by_la ?? {};
+  expect(Object.keys(averages).length).toBeGreaterThan(100);
+
+  type Row = { urn: number; school_type?: string; local_authority?: string; attainment_8_score?: number | null };
+  const compared = (s: Row) => s.attainment_8_score != null && s.local_authority != null
+    && averages[s.local_authority] != null
+    && !/special|pupil referral|alternative provision/i.test(s.school_type ?? '');
+  let found: { la: string; state: Row; independent: Row } | null = null;
+  for (const name of ['Kensington and Chelsea', 'Westminster', 'Camden', 'Hammersmith and Fulham', 'Barnet']) {
+    // The search page asks for the same first 50 schools.
+    const res = await page.request.get(`/api/schools?search=${encodeURIComponent(name)}&phase=secondary&page_size=50`);
+    const schools: Row[] = ((await res.json()).schools ?? []).filter(compared);
+    const independent = schools.find(s => /independent/i.test(s.school_type ?? ''));
+    const state = schools.find(s => !/independent/i.test(s.school_type ?? ''));
+    if (independent && state) { found = { la: name, state, independent }; break; }
+  }
+  test.skip(!found, 'no LA here lists a state and an independent secondary on one page');
+
+  await page.goto(`/?search=${encodeURIComponent(found!.la)}&phase=secondary`);
+  const stats = (urn: number) => page.locator(`a[href^="/school/${urn}-"]`).first()
+    .locator('xpath=ancestor::div[contains(@class, "__rowContent")][1]')
+    .locator('[class*="__line3"]');
+  await expect(stats(found!.state.urn).getByText(/vs LA avg/)).toBeVisible({ timeout: 15_000 });
+  const independent = stats(found!.independent.urn);
+  await expect(independent.getByText(found!.independent.attainment_8_score!.toFixed(1))).toBeVisible();
+  await expect(independent.getByText(/vs LA avg/)).toHaveCount(0);
+});
+
+test('a secondary shows its 2023/24 GCSE results, the last year DfE published Progress 8', async ({ page }) => {
+  // DfE re-issued its 2023/24 file under older column names, and every
+  // school's 2023/24 row loaded empty (audit C2). These are DfE's final
+  // figures for Bishop Stopford School, so they do not change.
+  await page.goto('/school/137086');
+  const history = page.locator('#history');
+  await history.getByText('View raw year-by-year data').click();
+  const row = history.getByRole('row', { name: /2023\/24/ });
+  await expect(row).toContainText('64.1');
+  await expect(row).toContainText('+1.0');
+  await expect(row).toContainText('91.7%');
+});
+
 test('a phase outside primary/secondary filters to that phase, not to everything', async ({ page }) => {
   // The search page offers every GIAS phase, but the API only knew the grouped
   // ones and silently dropped the rest — so "Nursery" returned primaries.

nextjs-app/__tests__/components/LeafletMapInner.test.tsx

@@ -105,3 +105,20 @@ it('puts the card back when the pins are rebuilt for a reason other than the sch
   expect(container.querySelector('.sc-popup')).toHaveTextContent('Southmead Primary School');
   expect(container.querySelectorAll('.sc-pin--selected')).toHaveLength(1);
 });
+
+it('compares a state secondary with its LA average on the card, but not an independent one', () => {
+  const state: School = { ...base, urn: 3, school_name: 'Holland Park School', school_type: 'Academy converter',
+    phase: 'Secondary', local_authority: 'Kensington and Chelsea', attainment_8_score: 60,
+    latitude: 51.5, longitude: -0.2, distance: 0.3 };
+  const independent: School = { ...state, urn: 4, school_name: 'Abbey Gate College',
+    school_type: 'Other independent school', attainment_8_score: 20.4 };
+  const laAverages = { 'Kensington and Chelsea': 54.5 };
+
+  const { container, rerender } = renderMap({ schools: [state, independent], laAverages, selectedUrn: 3 });
+  expect(container.querySelector('.sc-popup')).toHaveTextContent('60.0 Att 8 +5.5 vs LA');
+
+  rerender({ schools: [state, independent], laAverages, selectedUrn: 4 });
+  const card = container.querySelector('.sc-popup')!;
+  expect(card).toHaveTextContent('20.4 Att 8');
+  expect(card).not.toHaveTextContent(/vs LA/);
+});

nextjs-app/__tests__/components/SecondarySchoolRow.test.tsx

@@ -116,3 +116,38 @@ describe('SecondarySchoolRow shares the school page flags', () => {
     expect(screen.getByText('Fee-paying')).toBeInTheDocument();
   });
 });
+
+describe('SecondarySchoolRow LA comparison', () => {
+  it('compares a state school with its LA average', () => {
+    render(
+      <SecondarySchoolRow
+        school={{ ...base, school_type: 'Academy converter', attainment_8_score: 60 }}
+        laAvgAttainment8={54.5}
+      />,
+    );
+    expect(screen.getByText(/\+5\.5 vs LA avg/)).toBeInTheDocument();
+  });
+
+  it('keeps the comparison for a school whose type is unknown', () => {
+    render(
+      <SecondarySchoolRow
+        school={{ ...base, school_type: null, attainment_8_score: 60 } as unknown as School}
+        laAvgAttainment8={54.5}
+      />,
+    );
+    expect(screen.getByText(/\+5\.5 vs LA avg/)).toBeInTheDocument();
+  });
+
+  it("shows an independent school's Attainment 8 without an LA comparison", () => {
+    // DfE's LA average covers state-funded schools; an independent's
+    // Attainment 8 leaves out IGCSEs (audit H2).
+    render(
+      <SecondarySchoolRow
+        school={{ ...base, school_type: 'Other independent school', attainment_8_score: 20.4 }}
+        laAvgAttainment8={54.5}
+      />,
+    );
+    expect(screen.getByText('20.4')).toBeInTheDocument();
+    expect(screen.queryByText(/vs LA avg/)).not.toBeInTheDocument();
+  });
+});

nextjs-app/__tests__/lib/utils.test.ts

@@ -391,3 +391,20 @@ describe('singleSexLabel', () => {
     expect(singleSexLabel(undefined)).toBeNull();
   });
 });
+
+describe('isIndependentSchool', () => {
+  const { isIndependentSchool } = require('@/lib/utils');
+
+  it('matches both GIAS independent types', () => {
+    expect(isIndependentSchool({ school_type: 'Other independent school' })).toBe(true);
+    expect(isIndependentSchool({ school_type: 'Other independent special school' })).toBe(true);
+  });
+
+  it('does not match state-funded types or a missing type', () => {
+    for (const t of ['Academy converter', 'Community school', 'Free schools', 'Non-maintained special school']) {
+      expect(isIndependentSchool({ school_type: t })).toBe(false);
+    }
+    expect(isIndependentSchool({ school_type: null })).toBe(false);
+    expect(isIndependentSchool({})).toBe(false);
+  });
+});

nextjs-app/components/LeafletMapInner.tsx

@@ -9,7 +9,7 @@ import { useEffect, useRef, useState } from 'react';
 import L from 'leaflet';
 import 'leaflet/dist/leaflet.css';
 import type { School } from '@/lib/types';
-import { schoolUrl, isSpecialSchool, buildOfstedListBadge, listRwmValue } from '@/lib/utils';
+import { schoolUrl, isSpecialSchool, isIndependentSchool, buildOfstedListBadge, listRwmValue } from '@/lib/utils';
 
 interface LeafletMapInnerProps {
   schools: School[];
@@ -70,7 +70,7 @@ function metricHtml(school: School, { nationalAvgRwm, laAverages }: CardContext)
     const score = school.attainment_8_score;
     const laAvg = school.local_authority ? (laAverages?.[school.local_authority] ?? null) : null;
     let delta = '';
-    if (!special && laAvg != null) {
+    if (!special && !isIndependentSchool(school) && laAvg != null) {
       const diff = Math.round((score - laAvg) * 10) / 10;
       // Att8 runs 0–90 in 0.1 steps; ±0.5 is meaningful, where RWM needs ±2.
       const cls = diff >= 0.5 ? 'sc-up' : diff <= -0.5 ? 'sc-down' : '';

nextjs-app/components/SecondarySchoolRow.tsx

@@ -11,7 +11,7 @@
 'use client';
 
 import type { School } from '@/lib/types';
-import { buildOfstedListBadge, getPhaseStyle, schoolUrl, formatAgeRange, isProposedToClose, isSpecialSchool } from '@/lib/utils';
+import { buildOfstedListBadge, getPhaseStyle, schoolUrl, formatAgeRange, isProposedToClose, isSpecialSchool, isIndependentSchool } from '@/lib/utils';
 import { schoolFlags, schoolTypeLabel } from '@/lib/schoolFacts';
 import styles from './SecondarySchoolRow.module.css';
 
@@ -45,10 +45,12 @@ export function SecondarySchoolRow({
   const att8 = school.attainment_8_score;
   // The school's own Attainment 8 is a same-school figure — shown whenever it
   // exists (special schools included; their type tag on line 2 gives context).
-  // Only the vs-LA-average delta, a benchmark comparison, is dropped for
+  // Only the vs-LA-average delta, a benchmark comparison, is dropped: for
-  // special schools / PRUs / AP, whose pupils aren't measured against it fairly.
+  // special schools / PRUs / AP, whose pupils aren't measured against it
+  // fairly, and for independent schools, because DfE's LA average covers
+  // state-funded schools and an independent's Attainment 8 leaves out IGCSEs.
   const laDelta =
-    att8 != null && !isSpecialSchool(school) && laAvgAttainment8 != null
+    att8 != null && !isSpecialSchool(school) && !isIndependentSchool(school) && laAvgAttainment8 != null
       ? att8 - laAvgAttainment8
       : null;
 

nextjs-app/lib/utils.ts

@@ -758,6 +758,16 @@ export function isSpecialSchool(school: { school_type?: string | null }): boolea
   return /\bspecial\b/.test(t) || /pupil referral/.test(t) || /alternative provision/.test(t);
 }
 
+/**
+ * Independent (fee-paying) schools: GIAS types "Other independent school" and
+ * "Other independent special school". DfE's Attainment 8 for them leaves out
+ * IGCSEs, and DfE's LA averages cover state-funded schools only, so callers
+ * drop the "vs LA avg" comparison for them (audit H2).
+ */
+export function isIndependentSchool(school: { school_type?: string | null }): boolean {
+  return /\bindependent\b/i.test(school.school_type ?? '');
+}
+
 /**
  * Whether GIAS records a religious character. "None", "Does not apply" and
  * "Not applicable" are the register's ways of saying it has none; the place

pipeline/dags/school_data_pipeline.py

@@ -106,9 +106,12 @@ print(f'Validation passed: {{count}} GIAS rows')
         """,
     )
 
+    # Marts fed by annual EES staging models are rebuilt by the EES DAG, even
+    # when they join dim_school. Selecting them here fails in any database
+    # where that DAG hasn't run (pipeline/tests/test_dag_selectors.py).
     dbt_build = BashOperator(
         task_id="dbt_build",
-        bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_gias_establishments+ stg_gias_links+ gias_code_names+ --exclude int_ks2_with_lineage+ int_ks4_with_lineage+",
+        bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_gias_establishments+ stg_gias_links+ gias_code_names+ --exclude int_ks2_with_lineage+ int_ks4_with_lineage+ stg_ees_ks4_destinations+ stg_ees_ks5_destinations+",
     )
 
     sync_typesense = BashOperator(
@@ -143,7 +146,7 @@ with DAG(
 
     dbt_build_ofsted = BashOperator(
         task_id="dbt_build",
-        bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_ofsted_inspections+ int_ofsted_latest+ fact_ofsted_inspection+ dim_school+",
+        bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_ofsted_inspections+ int_ofsted_latest+ fact_ofsted_inspection+ dim_school+ --exclude stg_ees_ks4_destinations+ stg_ees_ks5_destinations+",
     )
 
     sync_typesense_ofsted = BashOperator(
@@ -190,7 +193,7 @@ with DAG(
 
     dbt_build_ees = BashOperator(
         task_id="dbt_build",
-        bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_ees_ks2+ stg_legacy_ks2+ stg_ees_ks4+ stg_legacy_ks4+ stg_ees_census+ stg_ees_admissions+ stg_ees_ks2_national+ stg_ees_ks4_national+ stg_ees_ks4_destinations+ stg_ees_ks5_destinations+ stg_ees_ks4_destinations_national+ stg_ees_ks5_destinations_national+",
+        bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_ees_ks2+ stg_legacy_ks2+ stg_ees_ks4+ stg_legacy_ks4+ stg_ees_census+ stg_ees_admissions+ stg_ees_ks2_national+ stg_ees_ks4_national+ stg_ees_ks4_la+ stg_ees_ks4_destinations+ stg_ees_ks5_destinations+ stg_ees_ks4_destinations_national+ stg_ees_ks5_destinations_national+",
     )
 
     sync_typesense_ees = BashOperator(

pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/ks4_info.py

@@ -0,0 +1,47 @@
+"""KS4 school information: the fields the stream declares, and the older
+names DfE used for them.
+
+2023/24 information exists only in the 2023/24 release, whose
+202324_information_about_schools_final.csv (re-issued 10 March 2026) uses the
+older names. Without the renames every 2023/24 field loaded as null (audit C2).
+Newer files contain none of the older names, so the renames leave them alone.
+
+Free of the Singer SDK so CI's pytest can load it.
+"""
+
+# Declared Singer fields besides the required time_period and school_urn.
+KS4_INFO_FIELDS = (
+    "school_laestab",
+    "school_name",
+    "establishment_type_group",
+    "reldenom",
+    "admpol_pt",
+    "egender",
+    "agerange",
+    "allks_pupil_count",
+    "allks_boys_count",
+    "allks_girls_count",
+    "endks4_pupil_count",
+    "ks2_scaledscore_average",
+    "sen_with_ehcp_pupil_percent",
+    "sen_pupil_percent",
+    "sen_no_ehcp_pupil_percent",
+    "attainment8_diffn",
+    "progress8_diffn",
+    "progress8_banding",
+)
+
+# 2023/24 column name → declared field.
+KS4_INFO_RENAMES = {
+    "t_allks_pupils": "allks_pupil_count",
+    "t_allks_boys": "allks_boys_count",
+    "t_allks_girls": "allks_girls_count",
+    "t_pupils": "endks4_pupil_count",
+    "avg_ks2_scaledscore": "ks2_scaledscore_average",
+    "pt_sen_with_ehcp": "sen_with_ehcp_pupil_percent",
+    "pt_sen": "sen_pupil_percent",
+    "pt_sen_no_ehcp": "sen_no_ehcp_pupil_percent",
+    "diffn_att8": "attainment8_diffn",
+    "diffn_p8mea": "progress8_diffn",
+    "p8_banding": "progress8_banding",
+}

pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/ks4_summary.py

@@ -0,0 +1,56 @@
+"""DfE's KS4 "summary, all state-funded" data set (Key stage 4 performance).
+
+One CSV holds England, regional and local-authority headline rows for every
+year since 2018/19. The England stream (ees_ks4_national) and the LA stream
+(ees_ks4_la) both read it. Its LA rows match DfE's published performance-table
+LA averages exactly (audit H2). Suppressed values ('z', 'x') become NULL in
+dbt; Progress 8 is 'z' in years with no KS2 baseline (2024/25): DfE policy,
+not missing data.
+
+Free of the Singer SDK so CI's pytest can load it.
+"""
+from __future__ import annotations
+
+import pandas as pd
+
+KS4_SUMMARY_CSV_URL = (
+    "https://explore-education-statistics.service.gov.uk/data-catalogue/"
+    "data-set/1b649e16-01e8-435b-a814-56be2faf9054/csv"
+)
+
+# CSV column → Singer field: the same 8 headline measures at every level.
+KS4_HEADLINE_COL_MAP = {
+    "attainment8_average":    "attainment_8_score",
+    "progress8_average":      "progress_8_score",
+    "engmath_94_percent":     "english_maths_standard_pass_pct",
+    "engmath_95_percent":     "english_maths_strong_pass_pct",
+    "ebacc_entering_percent": "ebacc_entry_pct",
+    "ebacc_94_percent":       "ebacc_standard_pass_pct",
+    "ebacc_95_percent":       "ebacc_strong_pass_pct",
+    "ebacc_aps_average":      "ebacc_avg_score",
+}
+
+
+def headline_rows(df: pd.DataFrame, geographic_level: str) -> pd.DataFrame:
+    """All-pupil rows for all state-funded schools at one geographic level
+    ("National" or "Local authority"). Column names are lower-cased first;
+    a filter column the file lacks is not applied."""
+    df = df.copy()
+    df.columns = [c.strip().lower() for c in df.columns]
+    for col, want in (
+        ("geographic_level", geographic_level),
+        ("establishment_type_group", "All state-funded"),
+        ("breakdown_topic", "Total"),
+        ("breakdown", "Total"),
+    ):
+        if col in df.columns:
+            df = df[df[col].str.strip().str.lower() == want.lower()]
+    return df
+
+
+def headline_record(row: pd.Series, keys: tuple[str, ...]) -> dict[str, str]:
+    """A Singer record: the identifying columns, then the headline measures."""
+    record = {key: str(row.get(key, "")).strip() for key in keys}
+    for csv_col, field in KS4_HEADLINE_COL_MAP.items():
+        record[field] = str(row.get(csv_col, "")).strip()
+    return record

pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/release_precedence.py

@@ -0,0 +1,59 @@
+"""Which release a row comes from when DfE re-publishes a year.
+
+DfE re-publishes earlier years inside later releases: the 2024/25 KS4 results
+file holds 2022/23, 2023/24 and 2024/25. The 2023/24 release's own file,
+re-issued on 10 March 2026 under older column names, was read after it and
+overwrote every 2023/24 row with blanks (audit C2). A stream that opts in
+treats the newest release as the authority for every year it contains.
+
+Free of the Singer SDK so CI's pytest, which installs only the backend's
+requirements, can load it.
+"""
+from __future__ import annotations
+
+import re
+
+import pandas as pd
+
+_SLUG_YEAR = re.compile(r"^(\d{4})-(\d{2})(?:-|$)")
+
+
+def slug_to_time_period(slug: str) -> str | None:
+    """A release slug's academic year as a time_period: '2022-23' → '202223'.
+
+    Suffixed slugs ('2024-25-revised', '2025-26-provisional') give the same
+    year, so newest_first places them by year; read as unknown, a revised
+    release went last and lost its year to the first release.
+    """
+    match = _SLUG_YEAR.match(slug or "")
+    return match.group(1) + match.group(2) if match else None
+
+
+def newest_first(releases: list[dict]) -> list[dict]:
+    """Releases by time_period, newest first. A release whose time_period is
+    unknown goes last, in the order given."""
+    dated = [r for r in releases if r.get("time_period")]
+    undated = [r for r in releases if not r.get("time_period")]
+    return sorted(dated, key=lambda r: r["time_period"], reverse=True) + undated
+
+
+def periods_in(df: pd.DataFrame) -> set[str]:
+    """The years a release's rows cover."""
+    if "time_period" not in df.columns:
+        return set()
+    return set(df["time_period"].astype(str).str.strip()) - {""}
+
+
+def drop_owned_periods(
+    df: pd.DataFrame, owned: set[str]
+) -> tuple[pd.DataFrame, dict[str, int]]:
+    """Drop the rows for years a newer release already supplied.
+
+    Returns the rows kept and, for each year dropped, how many rows went.
+    """
+    if "time_period" not in df.columns or not owned:
+        return df, {}
+    periods = df["time_period"].astype(str).str.strip()
+    dropped = periods.isin(owned)
+    skipped = {str(k): int(v) for k, v in periods[dropped].value_counts().items()}
+    return df[~dropped], skipped

pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py

@@ -17,6 +17,20 @@ import requests
 from singer_sdk import Stream, Tap
 from singer_sdk import typing as th
 
+from tap_uk_ees.ks4_info import KS4_INFO_FIELDS, KS4_INFO_RENAMES
+from tap_uk_ees.ks4_summary import (
+    KS4_HEADLINE_COL_MAP,
+    KS4_SUMMARY_CSV_URL,
+    headline_record,
+    headline_rows,
+)
+from tap_uk_ees.release_precedence import (
+    drop_owned_periods,
+    newest_first,
+    periods_in,
+    slug_to_time_period,
+)
+
 CONTENT_API_BASE = (
     "https://content.explore-education-statistics.service.gov.uk/api"
 )
@@ -31,14 +45,6 @@ def get_content_release_id(publication_slug: str) -> str:
     return resp.json()["id"]
 
 
-def _slug_to_time_period(slug: str) -> str | None:
-    """Convert a release slug like '2022-23' to a time_period like '202223'."""
-    parts = slug.split("-")
-    if len(parts) == 2 and len(parts[0]) == 4 and len(parts[1]) == 2:
-        return parts[0] + parts[1]
-    return None
-
-
 def get_all_releases(publication_slug: str) -> list[dict]:
     """Return all releases for a publication as dicts with 'id' and 'time_period'.
 
@@ -63,7 +69,7 @@ def get_all_releases(publication_slug: str) -> list[dict]:
             total_pages = paging.get("totalPages", 1)
 
         for r in releases:
-            time_period = _slug_to_time_period(r.get("slug", ""))
+            time_period = slug_to_time_period(r.get("slug", ""))
             result.append({"id": r["id"], "time_period": time_period})
 
         if page >= total_pages:
@@ -88,6 +94,8 @@ class EESDatasetStream(Stream):
     target CSV path inside the ZIP (substring match, not exact).
     Subclasses may set _column_renames to map messy CSV column names to
     clean Singer field names before yielding records.
+    Subclasses may set _newest_release_owns_period when DfE re-publishes
+    earlier years in later releases and the newest copy is the authority.
     """
 
     replication_key = None
@@ -96,6 +104,7 @@ class EESDatasetStream(Stream):
     _urn_column: str = "school_urn"  # column name for URN in the CSV
     _encoding: str = "utf-8"  # CSV file encoding (some DfE files use latin-1)
     _column_renames: dict = {}  # CSV column name → Singer field name
+    _newest_release_owns_period: bool = False  # see release_precedence.py
 
     def get_records(self, context):
         import pandas as pd
@@ -110,6 +119,9 @@ class EESDatasetStream(Stream):
         self.logger.info(
             "Found %d release(s) for %s", len(releases), self._publication_slug
         )
+        if self._newest_release_owns_period:
+            releases = newest_first(releases)
+        owned_periods: set[str] = set()
 
         for release in releases:
             release_id = release["id"]
@@ -163,6 +175,15 @@ class EESDatasetStream(Stream):
             if urn_col in df.columns:
                 df = df[df[urn_col].notna() & (df[urn_col] != "")]
 
+            if self._newest_release_owns_period:
+                df, skipped = drop_owned_periods(df, owned_periods)
+                for period, count in sorted(skipped.items()):
+                    self.logger.info(
+                        "Skipping %d rows for %s from release %s: a newer release supplied that year",
+                        count, period, release_id,
+                    )
+                owned_periods |= periods_in(df)
+
             self.logger.info("Emitting %d school-level rows from release %s", len(df), release_id)
 
             for _, row in df.iterrows():
@@ -251,6 +272,9 @@ class EESKS4PerformanceStream(EESDatasetStream):
     primary_keys = ["school_urn", "time_period", "breakdown_topic", "breakdown", "sex"]
     _publication_slug = "key-stage-4-performance"
     _target_filename = "performance_tables_schools"
+    # DfE's 2024/25 file re-publishes 2022/23 and 2023/24 under current names;
+    # the 2023/24 release's own file uses older ones (audit C2).
+    _newest_release_owns_period = True
     schema = th.PropertiesList(
         th.Property("time_period", th.StringType, required=True),
         th.Property("school_urn", th.StringType, required=True),
@@ -310,34 +334,20 @@ class EESKS4PerformanceStream(EESDatasetStream):
 
 
 # ── KS4 Information (wide format: one row per school, context/demographics) ──
-# File: 202425_information_about_schools_provisional.csv (38 cols)
+# Files: 202425_information_about_schools_final.csv (38 cols, current names);
+# 202324_information_about_schools_final.csv (60 cols, older names — the only
+# source of 2023/24 information). Field list and renames: ks4_info.py.
 
 class EESKS4InfoStream(EESDatasetStream):
     name = "ees_ks4_info"
     primary_keys = ["school_urn", "time_period"]
     _publication_slug = "key-stage-4-performance"
     _target_filename = "information_about_schools"
+    _column_renames = KS4_INFO_RENAMES
     schema = th.PropertiesList(
         th.Property("time_period", th.StringType, required=True),
         th.Property("school_urn", th.StringType, required=True),
-        th.Property("school_laestab", th.StringType),
+        *[th.Property(field, th.StringType) for field in KS4_INFO_FIELDS],
-        th.Property("school_name", th.StringType),
-        th.Property("establishment_type_group", th.StringType),
-        th.Property("reldenom", th.StringType),
-        th.Property("admpol_pt", th.StringType),
-        th.Property("egender", th.StringType),
-        th.Property("agerange", th.StringType),
-        th.Property("allks_pupil_count", th.StringType),
-        th.Property("allks_boys_count", th.StringType),
-        th.Property("allks_girls_count", th.StringType),
-        th.Property("endks4_pupil_count", th.StringType),
-        th.Property("ks2_scaledscore_average", th.StringType),
-        th.Property("sen_with_ehcp_pupil_percent", th.StringType),
-        th.Property("sen_pupil_percent", th.StringType),
-        th.Property("sen_no_ehcp_pupil_percent", th.StringType),
-        th.Property("attainment8_diffn", th.StringType),
-        th.Property("progress8_diffn", th.StringType),
-        th.Property("progress8_banding", th.StringType),
     ).to_dict()
 
 
@@ -564,37 +574,23 @@ class EESKs2NationalStream(Stream):
             yield record
 
 
-# ── KS4 National Headlines (national level only — one row per year) ──────────
+# ── KS4 National and LA Headlines (one data set, two streams) ────────────────
-# Dataset: "National characteristics summary data" (Key stage 4 performance).
+# DfE's "summary, all state-funded" data set: England, regional and LA rows,
-# Official England state-funded headline measures, 2018/19 → latest.
+# 2018/19 → latest. URL, measures and filters: ks4_summary.py.
-# Suppressed values ('z', 'x') → NULL downstream. Progress 8 is legitimately
-# absent in years with no KS2 baseline (e.g. 2024/25) — that is DfE policy,
-# not missing data.
 
-_KS4_NATIONAL_CSV_URL = (
+def _read_ks4_summary(logger):
-    "https://explore-education-statistics.service.gov.uk/data-catalogue/"
+    """Download DfE's KS4 summary data set."""
-    "data-set/1b649e16-01e8-435b-a814-56be2faf9054/csv"
+    import pandas as pd
-)
 
-_KS4_NATIONAL_COL_MAP = {
+    logger.info("Downloading KS4 summary data set: %s", KS4_SUMMARY_CSV_URL)
-    "attainment8_average":    "attainment_8_score",
+    resp = requests.get(KS4_SUMMARY_CSV_URL, timeout=60)
-    "progress8_average":      "progress_8_score",
+    resp.raise_for_status()
-    "engmath_94_percent":     "english_maths_standard_pass_pct",
+    return pd.read_csv(io.BytesIO(resp.content), dtype=str, keep_default_na=False)
-    "engmath_95_percent":     "english_maths_strong_pass_pct",
-    "ebacc_entering_percent": "ebacc_entry_pct",
-    "ebacc_94_percent":       "ebacc_standard_pass_pct",
-    "ebacc_95_percent":       "ebacc_strong_pass_pct",
-    "ebacc_aps_average":      "ebacc_avg_score",
-}
 
 
 class EESKs4NationalStream(Stream):
-    """National KS4 headline averages — one row per academic year.
+    """National KS4 headline averages — one row per academic year (England,
-
+    all state-funded schools, all pupils)."""
-    Filters to geographic_level == 'National', establishment_type_group ==
-    'All state-funded', breakdown_topic == 'Total', breakdown == 'Total'
-    so only the England-wide all-pupils row per year is emitted.
-    """
 
     name = "ees_ks4_national"
     primary_keys = ["time_period"]
@@ -602,34 +598,41 @@ class EESKs4NationalStream(Stream):
 
     schema = th.PropertiesList(
         th.Property("time_period", th.StringType, required=True),
-        *[th.Property(out, th.StringType) for out in _KS4_NATIONAL_COL_MAP.values()],
+        *[th.Property(out, th.StringType) for out in KS4_HEADLINE_COL_MAP.values()],
     ).to_dict()
 
     def get_records(self, context):
-        import pandas as pd
+        df = headline_rows(_read_ks4_summary(self.logger), "National")
-
-        self.logger.info("Downloading KS4 national headlines: %s", _KS4_NATIONAL_CSV_URL)
-        resp = requests.get(_KS4_NATIONAL_CSV_URL, timeout=60)
-        resp.raise_for_status()
-
-        df = pd.read_csv(io.BytesIO(resp.content), dtype=str, keep_default_na=False)
-        df.columns = [c.strip().lower() for c in df.columns]
-
-        for col, want in [
-            ("geographic_level", "national"),
-            ("establishment_type_group", "all state-funded"),
-            ("breakdown_topic", "total"),
-            ("breakdown", "total"),
-        ]:
-            if col in df.columns:
-                df = df[df[col].str.strip().str.lower() == want]
-
         self.logger.info("Emitting %d national KS4 rows", len(df))
         for _, row in df.iterrows():
-            record = {"time_period": row.get("time_period", "").strip()}
+            yield headline_record(row, ("time_period",))
-            for csv_col, field in _KS4_NATIONAL_COL_MAP.items():
+
-                record[field] = row.get(csv_col, "").strip()
+
-            yield record
+class EESKs4LaStream(Stream):
+    """DfE's KS4 local-authority averages — one row per academic year and LA
+    (all state-funded schools, all pupils), from the same data set as
+    ees_ks4_national. They match DfE's published performance-table LA averages
+    and replace a mean the API took over every school, independent and special
+    included (audit H2). old_la_code is the GIAS LA code: schools join on it,
+    not on the name."""
+
+    name = "ees_ks4_la"
+    primary_keys = ["time_period", "old_la_code"]
+    replication_key = None
+
+    schema = th.PropertiesList(
+        th.Property("time_period", th.StringType, required=True),
+        th.Property("old_la_code", th.StringType, required=True),
+        th.Property("new_la_code", th.StringType),
+        th.Property("la_name", th.StringType),
+        *[th.Property(out, th.StringType) for out in KS4_HEADLINE_COL_MAP.values()],
+    ).to_dict()
+
+    def get_records(self, context):
+        df = headline_rows(_read_ks4_summary(self.logger), "Local authority")
+        self.logger.info("Emitting %d LA KS4 rows", len(df))
+        for _, row in df.iterrows():
+            yield headline_record(row, ("time_period", "old_la_code", "new_la_code", "la_name"))
 
 
 # ── Legacy KS2 (pre-COVID wide format from DfE performance tables) ────────────
@@ -972,6 +975,7 @@ class TapUKEES(Tap):
             LegacyKS4Stream(self),
             EESKs2NationalStream(self),
             EESKs4NationalStream(self),
+            EESKs4LaStream(self),
         ]
 
 

pipeline/plugins/extractors/tap-uk-gias/tap_uk_gias/gias_csv.py

@@ -0,0 +1,26 @@
+"""Read a GIAS extract from the raw bytes of the download.
+
+GIAS writes its CSVs in Windows-1252 and sends no charset, so `resp.text`
+leaves requests to guess the codec. On 3 Oct 2026 it guessed windows-1250 and
+"à" became "ŕ". Decode the bytes ourselves instead.
+"""
+
+from __future__ import annotations
+
+import io
+
+import pandas as pd
+
+GIAS_ENCODING = "cp1252"
+
+
+def read_gias_csv(content: bytes, logger=None) -> pd.DataFrame:
+    """Every column as a string; a blank cell stays ''."""
+    # Windows-1252 leaves five bytes undefined. One stray byte must not stop
+    # the daily refresh of every school, so it becomes U+FFFD and is logged.
+    text = content.decode(GIAS_ENCODING, errors="replace")
+    undecodable = text.count("�")
+    if undecodable and logger is not None:
+        logger.warning("%d byte(s) in the GIAS extract could not be decoded as %s",
+                       undecodable, GIAS_ENCODING)
+    return pd.read_csv(io.StringIO(text), dtype=str, keep_default_na=False)

pipeline/plugins/extractors/tap-uk-gias/tap_uk_gias/tap.py

@@ -7,6 +7,8 @@ from datetime import date, timedelta
 from singer_sdk import Stream, Tap
 from singer_sdk import typing as th
 
+from tap_uk_gias.gias_csv import read_gias_csv
+
 GIAS_URL_TEMPLATE = (
     "https://ea-edubase-api-prod.azurewebsites.net"
     "/edubase/downloads/public/edubasealldata{date}.csv"
@@ -74,9 +76,6 @@ class GIASEstablishmentsStream(Stream):
 
     def get_records(self, context):
         """Download GIAS CSV and yield rows."""
-        import io
-
-        import pandas as pd
         import requests
 
         today = date.today()
@@ -94,12 +93,7 @@ class GIASEstablishmentsStream(Stream):
 
         resp.raise_for_status()
 
-        df = pd.read_csv(
+        df = read_gias_csv(resp.content, self.logger)
-            io.StringIO(resp.text),
-            encoding="latin-1",
-            dtype=str,
-            keep_default_na=False,
-        )
 
         for _, row in df.iterrows():
             record = row.to_dict()
@@ -126,9 +120,6 @@ class GIASLinksStream(Stream):
 
     def get_records(self, context):
         """Download GIAS links CSV and yield rows."""
-        import io
-
-        import pandas as pd
         import requests
 
         today = date.today()
@@ -146,12 +137,7 @@ class GIASLinksStream(Stream):
 
         resp.raise_for_status()
 
-        df = pd.read_csv(
+        df = read_gias_csv(resp.content, self.logger)
-            io.StringIO(resp.text),
-            encoding="latin-1",
-            dtype=str,
-            keep_default_na=False,
-        )
 
         for _, row in df.iterrows():
             record = row.to_dict()

pipeline/tests/test_dag_selectors.py

@@ -0,0 +1,98 @@
+"""Every scheduled dbt build must only build models whose parents exist.
+
+The daily GIAS build selects `stg_gias_establishments+`, so any mart that joins
+dim_school joins the daily build too. When such a mart also reads a staging
+model that only the manually triggered EES DAG builds, the daily build fails in
+any database where that DAG has not run since. Sync and cache invalidation then
+never run either. The destinations marts did this from late August 2026.
+
+The graph is read from the model SQL, because CI has no dbt.
+"""
+import re
+from collections import defaultdict
+from pathlib import Path
+
+import pytest
+
+PIPELINE = Path(__file__).resolve().parents[1]
+MODELS = PIPELINE / 'transform' / 'models'
+DAG_FILE = PIPELINE / 'dags' / 'school_data_pipeline.py'
+
+REF = re.compile(r"ref\(\s*'([a-z0-9_]+)'\s*\)")
+DBT_BUILD = re.compile(r'dbt_build\w*\s*=\s*BashOperator\(.*?build --profiles-dir \. --target production ([^"]+)"', re.S)
+DAG_ID = re.compile(r'dag_id="([a-z0-9_]+)"')
+
+DAILY = 'school_data_daily'
+
+# dim_school reads int_ofsted_latest only when the relation exists
+# (adapter.get_relation), so a missing table is not a failure.
+OPTIONAL_PARENTS = {'int_ofsted_latest'}
+
+
+def model_parents():
+    """{model: models it refs}. Seeds are left out: they are loaded once and always exist."""
+    sql = {p.stem: p.read_text() for p in MODELS.rglob('*.sql')}
+    return {name: set(REF.findall(text)) & set(sql) for name, text in sql.items()}
+
+
+def downstream(node, children):
+    seen, stack = {node}, [node]
+    while stack:
+        for child in children[stack.pop()]:
+            if child not in seen:
+                seen.add(child)
+                stack.append(child)
+    return seen
+
+
+def expand(tokens, children):
+    out = set()
+    for token in tokens:
+        out |= downstream(token[:-1], children) if token.endswith('+') else {token}
+    return out
+
+
+def scheduled_builds():
+    """{dag_id: dbt selection arguments} for every dbt build in the DAG file."""
+    text = DAG_FILE.read_text()
+    starts = [(m.start(), m.group(1)) for m in DAG_ID.finditer(text)]
+    builds = {}
+    for i, (start, dag_id) in enumerate(starts):
+        end = starts[i + 1][0] if i + 1 < len(starts) else len(text)
+        found = DBT_BUILD.search(text, start, end)
+        if found:
+            builds[dag_id] = found.group(1)
+    return builds
+
+
+def selected_models(args, parents):
+    children = defaultdict(set)
+    for model, ps in parents.items():
+        for p in ps:
+            children[p].add(model)
+    select = re.search(r'--select (.+?)(?= --exclude|$)', args).group(1).split()
+    excluded = re.search(r'--exclude (.+)$', args)
+    exclude = excluded.group(1).split() if excluded else []
+    return (expand(select, children) - expand(exclude, children)) & set(parents)
+
+
+PARENTS = model_parents()
+BUILDS = scheduled_builds()
+DAILY_MODELS = selected_models(BUILDS[DAILY], PARENTS)
+
+
+def test_every_dag_with_a_dbt_build_is_parsed():
+    assert set(BUILDS) == {
+        'school_data_daily', 'school_data_monthly_ofsted', 'school_data_annual_ees',
+        'school_data_annual_idaci', 'school_data_annual_distance',
+    }
+
+
+@pytest.mark.parametrize('dag_id', sorted(BUILDS))
+def test_selected_models_only_read_models_that_exist(dag_id):
+    selected = selected_models(BUILDS[dag_id], PARENTS)
+    # The daily build is the base layer: other DAGs may rely on what it builds.
+    available = selected | OPTIONAL_PARENTS | (DAILY_MODELS if dag_id != DAILY else set())
+    missing = {model: sorted(PARENTS[model] - available) for model in sorted(selected)
+               if PARENTS[model] - available}
+    assert missing == {}, f'{dag_id} builds models whose parents it never builds: {missing}'

pipeline/tests/test_ees_ks4_info_names.py

@@ -0,0 +1,67 @@
+"""KS4 school information for 2023/24 exists only in the 2023/24 release,
+whose file uses DfE's older column names. The stream declared only the newer
+ones, so every 2023/24 field loaded as null (audit C2). The headers below are
+DfE's, copied from 202324_information_about_schools_final.csv and
+202425_information_about_schools_final.csv.
+"""
+import importlib.util
+from pathlib import Path
+
+import pytest
+
+MODULE = (Path(__file__).resolve().parents[1] / 'plugins' / 'extractors' / 'tap-uk-ees'
+          / 'tap_uk_ees' / 'ks4_info.py')
+
+HEADER_2023_24 = (
+    'time_period', 'time_identifier', 'geographic_level', 'country_code', 'country_name',
+    'school_laestab', 'school_urn', 'school_name', 'old_la_code', 'new_la_code', 'la_name',
+    'version', 'establishment_type_group', 'full_address', 'telnum', 'pcon_code', 'pcon_name',
+    'contflag', 'iclose', 'reldenom', 'admpol_pt', 'egender', 'feeder', 'agerange',
+    't_allks_pupils', 't_allks_boys', 't_allks_girls', 't_pupils', 't_boys', 'pt_boys',
+    't_girls', 'pt_girls', 'avg_ks2_scaledscore', 't_prior_lo', 'pt_prior_lo', 't_prior_av',
+    'pt_prior_av', 't_prior_hi', 'pt_prior_hi', 't_disadvantaged', 'pt_disadvantaged',
+    't_not_disadvantaged', 'pt_not_disadvantaged', 't_language_not_english',
+    'pt_language_not_english', 't_language_english', 'pt_language_english',
+    't_language_unknown', 'pt_language_unknown', 't_not_mobile', 'pt_not_mobile',
+    't_sen_with_ehcp', 'pt_sen_with_ehcp', 't_sen', 'pt_sen', 't_sen_no_ehcp',
+    'pt_sen_no_ehcp', 'diffn_att8', 'diffn_p8mea', 'p8_banding',
+)
+
+HEADER_2024_25 = (
+    'time_period', 'time_identifier', 'geographic_level', 'country_code', 'country_name',
+    'school_laestab', 'school_urn', 'school_name', 'old_la_code', 'new_la_code', 'la_name',
+    'version', 'establishment_type_group', 'full_address', 'telnum', 'pcon_code', 'pcon_name',
+    'contflag', 'iclose', 'reldenom', 'admpol_pt', 'egender', 'feeder', 'agerange',
+    'allks_pupil_count', 'allks_boys_count', 'allks_girls_count', 'endks4_pupil_count',
+    'ks2_scaledscore_average', 'sen_with_ehcp_pupil_count', 'sen_with_ehcp_pupil_percent',
+    'sen_pupil_count', 'sen_pupil_percent', 'sen_no_ehcp_pupil_count',
+    'sen_no_ehcp_pupil_percent', 'attainment8_diffn', 'progress8_diffn', 'progress8_banding',
+)
+
+
+@pytest.fixture
+def ks4_info():
+    spec = importlib.util.spec_from_file_location('ks4_info', MODULE)
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+def test_every_declared_field_is_in_the_2023_24_file_once_renamed(ks4_info):
+    renamed = {ks4_info.KS4_INFO_RENAMES.get(c, c) for c in HEADER_2023_24}
+    assert set(ks4_info.KS4_INFO_FIELDS) - renamed == set()
+
+
+def test_every_declared_field_is_in_the_2024_25_file(ks4_info):
+    assert set(ks4_info.KS4_INFO_FIELDS) - set(HEADER_2024_25) == set()
+
+
+def test_the_renames_cannot_collide_with_either_file(ks4_info):
+    # An old name in the current file, or a new name already in the old file,
+    # would let a rename overwrite a real column.
+    assert set(ks4_info.KS4_INFO_RENAMES) & set(HEADER_2024_25) == set()
+    assert set(ks4_info.KS4_INFO_RENAMES.values()) & set(HEADER_2023_24) == set()
+
+
+def test_every_rename_names_a_declared_field(ks4_info):
+    assert set(ks4_info.KS4_INFO_RENAMES.values()) <= set(ks4_info.KS4_INFO_FIELDS)

pipeline/tests/test_ees_ks4_summary.py

@@ -0,0 +1,72 @@
+"""DfE's KS4 "summary, all state-funded" data set holds England, regional and
+LA rows. The England stream kept only the England row; its LA rows match
+DfE's published LA averages exactly, where the API's own mean was 7 points
+low (audit H2). Values below are DfE's for Kensington and Chelsea (207) and
+Wandsworth (212).
+"""
+import importlib.util
+import io
+from pathlib import Path
+
+import pandas as pd
+import pytest
+
+MODULE = (Path(__file__).resolve().parents[1] / 'plugins' / 'extractors' / 'tap-uk-ees'
+          / 'tap_uk_ees' / 'ks4_summary.py')
+
+CSV = """time_period,geographic_level,old_la_code,new_la_code,la_name,establishment_type_group,breakdown_topic,breakdown,attainment8_average,progress8_average,engmath_94_percent,engmath_95_percent,ebacc_entering_percent,ebacc_94_percent,ebacc_95_percent,ebacc_aps_average
+202425,National,,,,All state-funded,Total,Total,46.1,z,64.5,45.7,40.5,26.9,17.7,4.1
+202425,National,,,,All state-funded,Sex,Boys,44.1,z,62.0,43.0,38.0,24.0,16.0,3.9
+202425,Regional,,,,All state-funded,Total,Total,47.2,z,66.0,47.0,41.0,28.0,18.0,4.2
+202425,Local authority,207,E09000020,Kensington and Chelsea,All state-funded,Total,Total,54.5,z,77,61.4,45.6,32,26.6,4.89
+202425,Local authority,207,E09000020,Kensington and Chelsea,All state-funded,Sex,Girls,57.0,z,80,64.0,48.0,35,28.0,5.1
+202324,Local authority,207,E09000020,Kensington and Chelsea,All state-funded,Total,Total,54.5,0.29,76,60.0,44.0,31,25.0,4.8
+202425,Local authority,212,E09000032,Wandsworth,All state-funded,Total,Total,51.8,z,72,55.0,50.0,33,24.0,4.6
+202425,Local authority,212,E09000032,Wandsworth,Academies and free schools,Total,Total,52.0,z,73,56.0,51.0,34,25.0,4.7
+"""
+
+LA_KEYS = ('time_period', 'old_la_code', 'new_la_code', 'la_name')
+
+
+def _df():
+    return pd.read_csv(io.StringIO(CSV), dtype=str, keep_default_na=False)
+
+
+@pytest.fixture
+def summary():
+    spec = importlib.util.spec_from_file_location('ks4_summary', MODULE)
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+def test_la_rows_are_one_per_la_and_year(summary):
+    rows = summary.headline_rows(_df(), 'Local authority')
+    assert sorted(zip(rows['time_period'], rows['old_la_code'])) == [
+        ('202324', '207'), ('202425', '207'), ('202425', '212')]
+
+
+def test_an_la_record_carries_codes_name_and_the_headline_measures(summary):
+    rows = summary.headline_rows(_df(), 'Local authority')
+    row = rows[(rows['time_period'] == '202425') & (rows['old_la_code'] == '207')].iloc[0]
+    assert summary.headline_record(row, LA_KEYS) == {
+        'time_period': '202425', 'old_la_code': '207', 'new_la_code': 'E09000020',
+        'la_name': 'Kensington and Chelsea',
+        'attainment_8_score': '54.5', 'progress_8_score': 'z',
+        'english_maths_standard_pass_pct': '77', 'english_maths_strong_pass_pct': '61.4',
+        'ebacc_entry_pct': '45.6', 'ebacc_standard_pass_pct': '32',
+        'ebacc_strong_pass_pct': '26.6', 'ebacc_avg_score': '4.89',
+    }
+
+
+def test_the_england_rows_are_one_per_year(summary):
+    rows = summary.headline_rows(_df(), 'National')
+    assert list(rows['time_period']) == ['202425']
+    assert summary.headline_record(rows.iloc[0], ('time_period',))['attainment_8_score'] == '46.1'
+
+
+def test_column_names_and_labels_match_whatever_their_case(summary):
+    df = _df()
+    df.columns = [c.upper() for c in df.columns]
+    df['GEOGRAPHIC_LEVEL'] = df['GEOGRAPHIC_LEVEL'].str.upper()
+    assert len(summary.headline_rows(df, 'Local authority')) == 3

pipeline/tests/test_ees_release_precedence.py

@@ -0,0 +1,100 @@
+"""DfE re-publishes earlier years inside later KS4 releases.
+
+The 2024/25 results file holds 2022/23, 2023/24 and 2024/25 under current
+column names. The 2023/24 release's own file, re-issued in March 2026 under
+older names, was read after it and overwrote every 2023/24 row with blanks
+(audit C2). For the KS4 results stream the newest release owns every year it
+contains.
+"""
+import importlib.util
+import re
+from pathlib import Path
+
+import pandas as pd
+import pytest
+
+TAP_DIR = (Path(__file__).resolve().parents[1] / 'plugins' / 'extractors' / 'tap-uk-ees'
+           / 'tap_uk_ees')
+
+
+@pytest.fixture
+def precedence():
+    spec = importlib.util.spec_from_file_location(
+        'release_precedence', TAP_DIR / 'release_precedence.py')
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+def _release(period):
+    return {'id': f'release-{period}', 'time_period': period}
+
+
+def test_releases_are_taken_newest_first_whatever_order_the_api_gives(precedence):
+    releases = [_release('202223'), _release('202425'), _release(None), _release('202324')]
+    ordered = precedence.newest_first(releases)
+    assert [r['time_period'] for r in ordered] == ['202425', '202324', '202223', None]
+
+
+def test_a_year_a_newer_release_supplied_is_dropped_from_an_older_one(precedence):
+    newer = pd.DataFrame({'time_period': ['202425', '202324', '202223'], 'school_urn': ['1'] * 3})
+    older = pd.DataFrame({'time_period': ['202324', '202324', '201920'], 'school_urn': ['1', '2', '1']})
+
+    kept, skipped = precedence.drop_owned_periods(older, precedence.periods_in(newer))
+
+    assert list(kept['time_period']) == ['201920']
+    assert skipped == {'202324': 2}
+
+
+def test_periods_match_despite_surrounding_spaces(precedence):
+    owned = precedence.periods_in(pd.DataFrame({'time_period': [' 202324 ']}))
+    kept, skipped = precedence.drop_owned_periods(pd.DataFrame({'time_period': ['202324']}), owned)
+    assert owned == {'202324'}
+    assert kept.empty
+    assert skipped == {'202324': 1}
+
+
+def test_nothing_is_dropped_before_any_year_is_owned(precedence):
+    df = pd.DataFrame({'time_period': ['202324'], 'school_urn': ['1']})
+    kept, skipped = precedence.drop_owned_periods(df, set())
+    assert kept.equals(df)
+    assert skipped == {}
+
+
+def test_a_file_without_time_period_is_left_alone(precedence):
+    df = pd.DataFrame({'school_urn': ['1']})
+    kept, skipped = precedence.drop_owned_periods(df, {'202324'})
+    assert kept.equals(df)
+    assert skipped == {}
+    assert precedence.periods_in(df) == set()
+
+
+def test_only_the_ks4_results_stream_opts_in():
+    # A general rule would wipe KS2: the 2024/25 KS2 file holds 98,448 of the
+    # 955,956 rows the 2023/24 release has for 2023/24.
+    source = (TAP_DIR / 'tap.py').read_text()
+    opted_in = [chunk.split('(')[0] for chunk in source.split('\nclass ')[1:]
+                if re.search(r'_newest_release_owns_period\s*=\s*True', chunk)]
+    assert opted_in == ['EESKS4PerformanceStream']
+
+
+@pytest.mark.parametrize('slug, period', [
+    ('2024-25', '202425'),
+    ('2024-25-revised', '202425'),
+    ('2025-26-provisional', '202526'),
+    ('latest', None),
+    ('', None),
+])
+def test_a_release_slug_gives_its_year_whatever_its_suffix(precedence, slug, period):
+    # KS2 already publishes "2024-25-revised" and "2025-26-provisional". An
+    # unread suffix sent the release last, behind the provisional one for the
+    # same year, which then owned the year and dropped every revised row.
+    assert precedence.slug_to_time_period(slug) == period
+
+
+def test_within_a_year_the_api_order_is_kept(precedence):
+    # The API lists a year's revised release before its first release.
+    releases = [{'id': 'revised', 'time_period': '202526'},
+                {'id': 'first', 'time_period': '202526'},
+                {'id': 'older', 'time_period': '202425'}]
+    assert [r['id'] for r in precedence.newest_first(releases)] == ['revised', 'first', 'older']

pipeline/tests/test_gias_encoding.py

@@ -0,0 +1,66 @@
+"""GIAS publishes its extracts in Windows-1252 and declares no charset.
+
+The tap used to hand pandas `resp.text`, so requests guessed the codec.
+On 3 Oct 2026 it guessed windows-1250, and "St Thomas à Becket" was stored
+as "St Thomas ŕ Becket". The `encoding=` passed to read_csv did nothing,
+because the text was already decoded.
+"""
+import importlib.util
+import logging
+from pathlib import Path
+
+import pytest
+
+MODULE = (Path(__file__).resolve().parents[1] / 'plugins' / 'extractors' / 'tap-uk-gias'
+          / 'tap_uk_gias' / 'gias_csv.py')
+
+
+@pytest.fixture
+def gias_csv():
+    spec = importlib.util.spec_from_file_location('gias_csv', MODULE)
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+# Byte for byte as GIAS writes it: 0xE0 à, 0x92 ’, 0xE9 é, 0xB0 °, 0xE7 ç.
+EXTRACT = (
+    b'"URN","EstablishmentName","HeadLastName"\r\n'
+    b'"138950","St Thomas \xe0 Becket Catholic Secondary School","Smith"\r\n'
+    b'"100000","The Dean and Chapter of St Paul\x92s Cathedral","Pr\xe9vert"\r\n'
+    b'"140677","North Star 180\xb0","Fran\xe7ois"\r\n'
+    b'"100001","No head recorded",""\r\n'
+)
+
+
+def test_names_decode_as_windows_1252(gias_csv):
+    df = gias_csv.read_gias_csv(EXTRACT)
+    assert list(df['EstablishmentName']) == [
+        'St Thomas à Becket Catholic Secondary School',
+        'The Dean and Chapter of St Paul’s Cathedral',
+        'North Star 180°',
+        'No head recorded',
+    ]
+    assert list(df['HeadLastName']) == ['Smith', 'Prévert', 'François', '']
+
+
+def test_the_codec_requests_guessed_is_not_used(gias_csv):
+    # What the tap stored on 3 Oct: the same bytes read as windows-1250.
+    assert 'ŕ' in EXTRACT.decode('cp1250')
+    names = ' '.join(gias_csv.read_gias_csv(EXTRACT)['EstablishmentName'])
+    assert 'ŕ' not in names
+
+
+def test_values_stay_strings(gias_csv):
+    df = gias_csv.read_gias_csv(EXTRACT)
+    assert df.loc[0, 'URN'] == '138950'
+
+
+def test_a_byte_windows_1252_leaves_undefined_does_not_stop_the_load(gias_csv, caplog):
+    # 0x81 has no Windows-1252 character. One odd name must not block the daily
+    # refresh of every school, but it must be visible in the log.
+    extract = b'"URN","EstablishmentName"\r\n"100002","Odd \x81 Name"\r\n'
+    with caplog.at_level(logging.WARNING):
+        df = gias_csv.read_gias_csv(extract, logger=logging.getLogger('gias'))
+    assert df.loc[0, 'EstablishmentName'] == 'Odd � Name'
+    assert 'could not be decoded' in caplog.text

pipeline/transform/macros/safe_numeric.sql

@@ -3,7 +3,9 @@
   Casts a string column to numeric, treating any non-numeric value as NULL.
   Handles all EES suppression codes (z, c, x, q, u, etc.) without needing
   an explicit list — any string that doesn't look like a number becomes NULL.
+  A trailing percent sign is accepted: DfE's 2023/24 KS2 information file
+  writes percentages as "34%" (audit C2).
 #}
 {% macro safe_numeric(col) -%}
-    CASE WHEN {{ col }} ~ '^-?[0-9]+(\.[0-9]+)?$' THEN {{ col }}::numeric ELSE NULL END
+    CASE WHEN {{ col }} ~ '^-?[0-9]+(\.[0-9]+)?%?$' THEN rtrim({{ col }}, '%')::numeric ELSE NULL END
 {%- endmacro %}

pipeline/transform/models/marts/_marts_schema.yml

@@ -279,11 +279,24 @@ models:
         tests: [not_null, unique]
 
   - name: fact_ks4_national_averages
-    description: Computed national KS4 averages (means across state schools in our dataset — not official DfE figures) — one row per academic year
+    description: Official DfE KS4 national headline averages (England, all state-funded schools, all pupils) — one row per academic year
     columns:
       - name: year
         tests: [not_null, unique]
 
+  - name: fact_ks4_la_averages
+    description: Official DfE KS4 local-authority averages (all state-funded schools, all pupils) — one row per academic year and LA; la_code is the GIAS LA code
+    columns:
+      - name: year
+        tests: [not_null]
+      - name: la_code
+        tests: [not_null]
+      - name: la_name
+        tests: [not_null]
+    tests:
+      - unique:
+          column_name: "year || '-' || la_code"
+
   - name: fact_deprivation
     description: IDACI deprivation index — one row per URN
     columns:

pipeline/transform/models/marts/fact_ks4_la_averages.sql

@@ -0,0 +1,22 @@
+{{ config(materialized='table') }}
+
+-- Mart: OFFICIAL DfE KS4 local-authority averages — one row per academic year
+-- and LA (all state-funded schools, all pupils), from the same EES data set
+-- as fact_ks4_national_averages. Feeds "vs LA avg" on search rows and map
+-- cards, replacing a mean the API took over every school in the LA,
+-- independent and special included (audit H2). la_code is the GIAS LA code.
+
+select
+    year,
+    la_code,
+    la_name,
+    attainment_8_score,
+    progress_8_score,
+    english_maths_standard_pass_pct,
+    english_maths_strong_pass_pct,
+    ebacc_entry_pct,
+    ebacc_standard_pass_pct,
+    ebacc_strong_pass_pct,
+    ebacc_avg_score
+from {{ ref('stg_ees_ks4_la') }}
+order by year, la_code

pipeline/transform/models/staging/_stg_sources.yml

@@ -79,6 +79,9 @@ sources:
       - name: ees_ks4_national
         description: Official KS4 national headline averages from DfE EES data catalogue — one row per academic year
 
+      - name: ees_ks4_la
+        description: Official KS4 local-authority averages (all state-funded schools) from the same DfE EES data set as ees_ks4_national — one row per academic year and LA
+
       # Phonics: no school-level data on EES (only national/LA level)
 
       - name: fbit_finance

pipeline/transform/models/staging/stg_ees_ks2.yml

@@ -0,0 +1,22 @@
+version: 2
+
+unit_tests:
+  - name: stg_ees_ks2_reads_percentages_written_with_a_sign
+    description: >
+      DfE's 2023/24 KS2 information file writes percentages as "34%", and every
+      2023/24 percentage loaded as null (audit C2). Plain numbers still load;
+      suppression codes stay null. School 147411's 2023/24 figures are DfE's.
+    model: stg_ees_ks2
+    given:
+      - input: source('raw', 'ees_ks2_attainment')
+        rows:
+          - {school_urn: '147411', time_period: '202324', subject: 'Reading, writing and maths', breakdown_topic: 'All pupils', breakdown: 'Total', expected_standard_pupil_percent: '70'}
+          - {school_urn: '100000', time_period: '202425', subject: 'Reading, writing and maths', breakdown_topic: 'All pupils', breakdown: 'Total', expected_standard_pupil_percent: '79'}
+      - input: source('raw', 'ees_ks2_info')
+        rows:
+          - {school_urn: '147411', time_period: '202324', totpups: '818', telig: '112', ptfsm6cla1a: '34%', ptealgrp2: '56%', psenelk: '20%', psenele: 'c', ptmobn: '87.5%'}
+          - {school_urn: '100000', time_period: '202425', totpups: '792', telig: '105', ptfsm6cla1a: '32', ptealgrp2: 'x', psenelk: '14', psenele: '2', ptmobn: '87'}
+    expect:
+      rows:
+        - {urn: 147411, year: 202324, total_pupils: 818, rwm_expected_pct: 70, disadvantaged_pct: 34, eal_pct: 56, sen_support_pct: 20, sen_ehcp_pct: null, stability_pct: 87.5}
+        - {urn: 100000, year: 202425, total_pupils: 792, rwm_expected_pct: 79, disadvantaged_pct: 32, eal_pct: null, sen_support_pct: 14, sen_ehcp_pct: 2, stability_pct: 87}

pipeline/transform/models/staging/stg_ees_ks4_la.sql

@@ -0,0 +1,21 @@
+-- Staging model: official DfE KS4 local-authority averages — one row per
+-- academic year and LA (all state-funded schools, all pupils). Same EES data
+-- set as stg_ees_ks4_national. la_code is the GIAS LA code
+-- (dim_location.local_authority_code). Suppressed values ('z', 'x') are
+-- coerced to NULL by safe_numeric.
+
+select
+    cast(trim(time_period) as integer)                        as year,
+    cast(trim(old_la_code) as integer)                        as la_code,
+    trim(la_name)                                             as la_name,
+    {{ safe_numeric('attainment_8_score') }}                  as attainment_8_score,
+    {{ safe_numeric('progress_8_score') }}                    as progress_8_score,
+    {{ safe_numeric('english_maths_standard_pass_pct') }}     as english_maths_standard_pass_pct,
+    {{ safe_numeric('english_maths_strong_pass_pct') }}       as english_maths_strong_pass_pct,
+    {{ safe_numeric('ebacc_entry_pct') }}                     as ebacc_entry_pct,
+    {{ safe_numeric('ebacc_standard_pass_pct') }}             as ebacc_standard_pass_pct,
+    {{ safe_numeric('ebacc_strong_pass_pct') }}               as ebacc_strong_pass_pct,
+    {{ safe_numeric('ebacc_avg_score') }}                     as ebacc_avg_score
+from {{ source('raw', 'ees_ks4_la') }}
+where time_period ~ '^[0-9]+$'
+  and old_la_code ~ '^[0-9]+$'

pipeline/transform/models/staging/stg_ees_ks4_la.yml

@@ -0,0 +1,15 @@
+version: 2
+
+unit_tests:
+  - name: stg_ees_ks4_la_casts_codes_years_and_measures
+    description: DfE's LA rows, keyed by the GIAS LA code. Suppressed values ('z') become null.
+    model: stg_ees_ks4_la
+    given:
+      - input: source('raw', 'ees_ks4_la')
+        rows:
+          - {time_period: '202425', old_la_code: '207', new_la_code: 'E09000020', la_name: 'Kensington and Chelsea', attainment_8_score: '54.5', progress_8_score: 'z', english_maths_standard_pass_pct: '77'}
+          - {time_period: '202324', old_la_code: '207', new_la_code: 'E09000020', la_name: 'Kensington and Chelsea', attainment_8_score: '54.5', progress_8_score: '0.29', english_maths_standard_pass_pct: '76'}
+    expect:
+      rows:
+        - {year: 202425, la_code: 207, la_name: 'Kensington and Chelsea', attainment_8_score: 54.5, progress_8_score: null, english_maths_standard_pass_pct: 77}
+        - {year: 202324, la_code: 207, la_name: 'Kensington and Chelsea', attainment_8_score: 54.5, progress_8_score: 0.29, english_maths_standard_pass_pct: 76}

pipeline/transform/tests/assert_ks2_info_percentages_loaded.sql

@@ -0,0 +1,16 @@
+-- Where a year's KS2 school information loaded (pupil counts present), its
+-- percentages loaded too. DfE's files give a disadvantaged % for 96–97% of
+-- those schools; 2023/24 loaded none because the file writes "34%" (audit
+-- C2). A year without an information file (2022/23) has no pupil counts and
+-- is skipped.
+
+select
+    year,
+    count(total_pupils) as with_pupils,
+    count(*) filter (where total_pupils is not null and disadvantaged_pct is not null)
+        as with_disadvantaged
+from {{ ref('stg_ees_ks2') }}
+group by year
+having count(total_pupils) >= 1000
+   and count(*) filter (where total_pupils is not null and disadvantaged_pct is not null)
+       < 0.9 * count(total_pupils)

pipeline/transform/tests/assert_ks4_la_averages_cover_las.sql

@@ -0,0 +1,13 @@
+-- DfE publishes an average for about 152 LAs a year. Fewer than 145 in the
+-- latest year (or none at all) means the LA filter in tap_uk_ees
+-- (ks4_summary.headline_rows) stopped matching, e.g. after DfE renamed a label.
+
+with latest as (
+    select max(year) as year from {{ ref('fact_ks4_la_averages') }}
+)
+
+select l.year, count(f.la_code) as las
+from latest l
+left join {{ ref('fact_ks4_la_averages') }} f on f.year = l.year
+group by l.year
+having count(f.la_code) < 145

pipeline/transform/tests/assert_ks4_la_averages_current.sql

@@ -0,0 +1,20 @@
+{{ config(severity='warn') }}
+
+-- DfE's LA averages cover the latest year with school results. They come from
+-- a fixed EES data set (ks4_summary.KS4_SUMMARY_CSV_URL); when DfE publishes a
+-- new year under a new data set, the school results move on and this mart does
+-- not, so /api/la-averages serves an empty map and every "vs LA avg" goes.
+-- A warning, not an error: it must not hold back a new year's school results.
+
+with schools as (
+    select max(year) as year from {{ ref('stg_ees_ks4') }} where attainment_8_score is not null
+),
+
+las as (
+    select max(year) as year from {{ ref('fact_ks4_la_averages') }}
+)
+
+select s.year as latest_school_year, l.year as latest_la_year
+from schools s
+cross join las l
+where l.year is null or l.year < s.year

pipeline/transform/tests/assert_ks4_years_have_results.sql

@@ -0,0 +1,12 @@
+-- Every year loaded from EES has an Attainment 8 for at least half its
+-- schools. DfE's files reach 82% each year; 2023/24 loaded at 0% after DfE
+-- re-issued its file under older column names (audit C2). Pre-2019 years come
+-- from stg_legacy_ks4 and are not checked here.
+
+select
+    year,
+    count(*) as schools,
+    count(attainment_8_score) as with_attainment_8
+from {{ ref('stg_ees_ks4') }}
+group by year
+having count(attainment_8_score) < 0.5 * count(*)