Measurement honesty so we optimise SAP-relevant accuracy, not SAP-neutral
misses (ADR-0030 Component Accuracy):
- Add construction_age_band_pm1: an exact-or-adjacent-band hit. Adjacent
RdSAP age bands carry near-identical U-values, so an off-by-one is
~SAP-neutral. Full corpus: exact 78.5% but ±1-band 91.7% (fixture
63.9% -> 83.3%) — most age misses are adjacent.
- Drop window_count from the gate's residual ceilings (cosmetic): the
predicted picture clusters at a mapper-default 4 windows vs actuals 1-21,
but total_window_area (the SAP-relevant signal) stays tight at ~3.4 m2.
Gate: + construction_age_band_pm1 floor 0.8333; window_count no longer gated.
Closes#1222
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Investigated recency-weighting (weight cohort votes by an exponential decay
in cert age). Key finding: it must be SELECTIVE. On the validation corpus it
HURTS permanent categoricals (wall 91.2->89.5, age 78.5->75.7 — discards
still-valid data) but clearly HELPS time-varying ones, where a recent
neighbour reflects the current physical state:
roof_insulation_thickness 56.7 -> 60.7% corpus (+4pp)
29.4 -> 41.2% fixture (+12pp)
So apply a recency-weighted mode only to roof_insulation_thickness (loft
top-ups happen over time); keep the plain mode for permanent categoricals.
tau = 4yr (~2.8yr half-life); falls back to plain mode when no registration
dates are lodged. Gate floor ratcheted 0.2941 -> 0.4118.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
These independent fabric categoricals were template-copied; mode them like
the construction categoricals. Verified mode beats template before applying.
Big fixture win on roof insulation thickness (doubled), floor insulation
neutral-to-positive:
roof_insulation_thickness 14.7% -> 29.4% (gate floor ratcheted up)
floor_insulation 90.6% (unchanged on the fixture)
Glazing type was tried too (+1.6pp on the 40-postcode corpus) but REGRESSED
the 36-target fixture (0.50 -> 0.44) — the gate caught it. Glazing moding is
marginal/noisy, so it's left on the template; revisit with a larger corpus.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Tier-2 (full national bulk streaming) is deferred. The near-term scale
validation is a Tier-1.5: a few-thousand-cert anonymised corpus stored in
S3 (too large to commit, far more stable than the 36-target gate fixture),
pulled to a temp dir and run through the same load_corpus +
evaluate_component_accuracy. Reuses the committed-fixture machinery wholesale
— only the data source differs. One scorer, three data sources (committed
fixture / S3 corpus / bulk stream).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The committed CI gate: run the calculator-free leave-one-out scorer over the
frozen anonymised fixture (36 SAP-10.2 targets) and assert each per-component
classification rate / geometry residual is no worse than a committed baseline.
Prediction is deterministic + the fixture frozen, so the numbers reproduce
exactly — a failure is a real regression, never sample noise.
- 19 rate floors + 5 residual ceilings, seeded at the currently-measured
values; they only ever tighten (no-widening ethos on an aggregate).
- Calculator-FREE — component floors are the real gate; the end-to-end
SAP/carbon/PE guards stay out (their floor is the separate API-path
calculator workstream).
- Skips with a message when the fixture is absent.
25 parametrized assertions, all green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The committed gate needs frozen, reproducible data without dumping real UK
addresses into the repo. Add:
- harness anonymise_payload + stable_hash: hash street address + cert number
into opaque, dedup-stable tokens; blank secondary address lines + post_town;
keep postcode + all component/lodged fields (gov data is OGL). Unit-tested.
- scripts/build_epc_prediction_fixture.py: curate qualifying postcodes (>=1
SAP 10.2 target + >=2 distinct addresses) from the local scratch corpus,
anonymise, freeze under tests/fixtures/epc_prediction/.
- The frozen fixture: 15 postcodes / 280 certs / 36 SAP-10.2 targets.
Verified no plaintext address_line_1 and post_town all blank.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
"One scorer, two harnesses" (ADR-0030): the committed gate, the local script,
and the future battle-test must run the *same* scoring. Extract it:
- domain/epc_prediction/validation.py — `iter_predictions` (the single
leave-one-out orchestration: latest-per-address hold-out, SAP-10.2 target
filter, all-vintage source) + `evaluate_component_accuracy` (calculator-free
ComponentAccuracy aggregation, the primary signal). Unit-tested.
- harness/epc_prediction_corpus.py — `load_corpus(dir)` IO: corpus dir ->
Comparable cohorts (maps payloads, carries address + registration_date).
validate_epc_prediction.py now just loads + calls the scorer for the component
section and iterates iter_predictions for the calculator-floored end-to-end.
Identical numbers (181 targets, SAP MAE 6.34) — behaviour-preserving.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Make the leave-one-out runner ADR-0030-compliant:
- Hold out only SAP 10.2 targets (sap_version == 10.2) — the source cohort
keeps every vintage (components are methodology-agnostic).
- Label Component Accuracy as the PRIMARY, calculator-independent section.
- End-to-end vs API-lodged (SECONDARY, calculator-FLOORED): add CO2 (tonnes)
and PEI (kWh/m2) alongside SAP, using the canonical performance.py mapping
(co2_kg/1000; primary_energy_kwh_per_m2).
- Add the attribution readout calc(actual) vs lodged SAP — the calculator
floor the end-to-end can reach.
- Drop the neighbour-mean-of-lodged-SAP baseline (mixes SAP versions —
rejected by ADR-0030).
On the 181 SAP-10.2 targets: component rates are higher than the all-vintage
view (age band 60.9 -> 78.5%, floor_area mean|.| 12.7 -> 8.4). End-to-end SAP
MAE 6.34 vs the calc(actual) floor of 3.25 — ~half the gap is the known
API-path calculator residual, not prediction error.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Heating is the dominant SAP lever (ablating it to actual cut the SAP error
~7 -> ~4.5) yet was entirely unscored. Add the heating group to
compare_prediction's categorical_hits: main fuel / category / control (off
the primary MainHeatingDetail), water-heating fuel / code, has-cylinder,
cylinder insulation, secondary heating (off SapHeating).
Template-copied baseline on the 40-postcode corpus (no predictor change
yet — this just makes the signal visible):
heating_main_fuel 93.4%
heating_main_category 92.7%
water_heating_fuel/code 91.7% / 92.4%
heating_main_control 62.1% <- weak
has_hot_water_cylinder 78.5%
cylinder_insulation_type 35.8% (n=120) <- weak
secondary_heating_type 16.8% (n=125) <- weak
Fuel/category predict well from the template; controls, cylinder, and
secondary heating are poor and now drive the next predictor slices.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
ADR-0030 commits Component Accuracy to ~19 categorical components (5 today
+ 8 heating + glazing/renewables). Flat *_correct dataclass fields don't
scale — each needs manual runner wiring. Collapse them into a single
`categorical_hits: dict[str, Optional[bool]]` keyed by component name, which
also matches the runner's name-keyed aggregation (now generic: it tallies
whatever components the comparison reports). No behaviour change; the
classification rates are identical (wall n 578->575 is the 3 certs whose
actual wall is None, now correctly counted as not-applicable via _classify).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Records the grilling-session decisions amending ADR-0029's validation:
- Source cohort keeps all cert vintages (components are agnostic of the SAP
methodology that rated them); only the held-out validation TARGET is
restricted to SAP 10.2. Amends ADR-0029 decision 5 ("pre-SAP10 dropped").
- Component Accuracy (predicted vs API actual components) is the primary,
calculator-independent signal. calc(predicted) vs calc(actual) rejected
(circular ground truth, hides calculator error); neighbour-mean-lodged-SAP
baseline rejected (mixes SAP versions). calc(predicted) vs API-lodged
SAP/carbon/PE kept as a secondary, calculator-floored guard.
- Two tiers: committed anonymized fixture (ratcheting CI gate) + bulk-export
national battle-test on harness/epc_bulk.py + harness/cohort.py, emitting
accuracy + a failure taxonomy, re-baselining the gate floors.
CONTEXT.md: Comparable Properties corrected to all-vintage source; new
Component Accuracy term. ADR-0029 Validation section marked superseded.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bump N_POSTCODES 150 -> 40 as the gradual-growth step from the 3-postcode
smoke. 40 postcodes / 1113 certs / 578 leave-one-out predictions is enough
for stable, trustworthy metrics (the smoke's 2 usable postcodes were
dominated by oddball flats — floor_area mean|.| 52.6 there vs 12.7 here).
Resumable + reproducible (random.seed(2026)); raise again to scale up.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The register lists every historical lodgement, so a postcode cohort
contains the same physical address many times (LS61AA: 15 certs / 11
addresses; NG71AA: 15 / 9 — "FLAT 3" appears 3x in each). Two
consequences:
- Production: a re-lodged neighbour was counting up to 3x towards the
cohort mode. select_comparables now dedupes candidates to the latest
cert per address (one comparable per real neighbour) — Comparable
gains address + registration_date (the register metadata its docstring
already anticipated, read straight off the cached payload).
- Validation: leave-one-out leaked — predicting a flat from a near-
identical re-lodgement of itself. The harness now holds out a whole
address (excludes every sibling cert) and evaluates on the latest cert
per address (the best ground truth).
Removing the leak gives the honest numbers (19 distinct addresses):
wall_construction 93.1% -> 89.5%
construction_age_band 65.5% -> 52.6%
roof_construction 79.3% -> 68.4%
floor_area mean|.| 37.9 -> 52.6 m2
The earlier figures were inflated by self-leakage; these are the real
accuracy to beat.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Only main wall_construction was set to the cohort mode; the other
homogeneous categoricals (wall insulation, construction age band, roof
construction, floor construction) were left as template-copied, so one
median-size template's quirks set them. Apply the same cohort-mode
mechanism to all of them per ADR-0029 decision 4 — the template still
supplies geometry, only the categorical codes move to the mode.
Verified mode beats (or ties) template-copy per categorical before
applying. Smoke corpus (29 leave-one-out) classification rates:
construction_age_band 55.2% -> 65.5%
roof_construction 72.4% -> 79.3%
floor_construction 46.2% -> 84.6%
wall_insulation_type 93.1% (tie — already template-strong)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The comparison only scored main wall_construction; everything else the
predictor produces (by template-copy) went unmeasured. Extend
compare_prediction to the rest of the ADR-0029 homogeneous categoricals —
wall insulation type, construction age band, roof construction, floor
construction — and aggregate per-categorical classification rates in the
runner. A categorical hit is "not applicable" (None, excluded from the
denominator) when the actual lodges no value, so absent-roof flats don't
score free wins.
Smoke corpus (29 leave-one-out, all but wall are template-copied today):
wall_construction 93.1%
wall_insulation_type 93.1%
construction_age_band 55.2% <- loud; candidate for cohort-mode
roof_construction 72.4%
floor_construction 46.2% (n=13)
These numbers drive the next slice (extend cohort-mode coverage).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Template (the comparable whose structure/geometry is copied wholesale)
was members[0] — an arbitrary draw from the API search order. With floor
area varying widely within a property_type cohort (NG71AA houses span
51-340 m2), this made the copied geometry noisy and systematically large.
Pick the member whose floor area is closest to the cohort median instead,
implementing ADR-0029 decision 4's unimplemented "closest on size"
criterion while keeping the structure coherent (it is still one real
property, so floor dims / windows / parts stay internally consistent for
the calculator).
Smoke corpus (29 leave-one-out predictions):
floor_area mean|.| 68.0 -> 37.9 m2 (bias +46.8 -> -3.9)
window_area mean|.| 11.1 -> 7.3 m2
parts mean|.| 1.00 -> 0.38
SAP |pred-calc - calc(actual)| MAE 7.19 -> 4.86
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Pure compare_prediction (TDD): wall-construction classification hit + signed
residuals on floor area, window count, total window area, building-parts count.
Plus validate_epc_prediction.py (IO plumbing): drops each cert from its postcode
cohort, predicts from the rest on guaranteed inputs only, aggregates the metrics,
and reports SAP three ways (pred-calc vs lodged / vs calc-on-actual / vs the
neighbour-mean baseline). Smoke run: wall 90.9%, floor-area mean|·| 42.6 m2 (a
real signal — template-copied floor area is noisy), SAP pred-calc edges baseline.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
predict() copies a representative template comparable's structure (coherent for
the calculator), overrides the homogeneous categorical with the cohort mode
(robust to an atypical template), then applies known Landlord Overrides on top
(a known value wins over the estimate). Proven on wall construction; roof/floor/
insulation/age extend on the same mode+override mechanism, driven next by the
validation harness metrics.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Pure-domain select_comparables: property type is an always-hard filter; built
form and known Landlord Overrides (e.g. solid brick) are conditioning filters on
the filter-then-relax ladder — applied while >= minimum_cohort survive, relaxed
otherwise (the mixed-street border case degrades gracefully). PredictionTarget
(known inputs) + Comparable (epc + register metadata) + ComparableProperties
(selected cohort). Weighting (recency x similarity) follows in the synthesis slice.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Builds the frozen validation corpus: samples postcodes from the register, then
caches each postcode's full cohort of raw cert payloads (the shape
from_api_response consumes), grouped by postcode, resumably. Reads the token
from backend/.env; cache dir /tmp/epc_prediction_corpus (EPC_PREDICTION_CORPUS
override). IO plumbing, not test-driven. Pairs with the leave-one-out harness.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Promote RdSAP-Schema-17.0 into SUPPORTED so all 1000 corpus certs are held to
the strict parse+map guard. Drop the now-redundant cert[0] tracer (subsumed by
the parametrised bucket); keep the reduced-field synthesis behavioural test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the 17.0 synthesis seam over the shared _synthesise_reduced_field_windows
core (inherited 20.0.0 coefficients, ND glazing -> DG-modal default 2, per
ADR-0028). 17.0 glazed_type codes (1-4,7) are a subset of the verified 1-8
space. The 10 rich certs use lodged window_area directly; the windowless 990
synthesise a 4-way N/E/S/W split.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Dispatch RdSAP-Schema-17.0 through from_api_response, parse-fix the schema
(data-driven required->optional, validated against the 1000-cert 17.0 corpus
per ADR-0028 — incl. SapHeating.cylinder_insulation_type and the
has_hot_water_cylinder / has_fixed_air_conditioning / has_heated_separate_
conservatory flags), and port the defensive mapper reads (dwelling_type
str/dict/number, photovoltaic_supply guard, sap_floor_dimensions guard). All
1000 corpus certs now parse and map.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reusable per-schema profiler: glazed_area band mix, Validation Cohort size,
observed-vs-predicted band glazing/floor ratio, and the ND/str sentinels that
drive schema widening. Regenerates the ADR-0028 transfer-check table from any
harvested corpus.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Promote RdSAP-Schema-19.0 into SUPPORTED so all 1000 corpus certs are held to
the strict parse+map guard. Drop the now-redundant cert[0] tracer (subsumed by
the parametrised bucket); keep the reduced-field synthesis behavioural test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the 19.0 synthesis seam over the shared _synthesise_reduced_field_windows
core (inherited 20.0.0 coefficients, ND glazing -> DG-modal default 2, per
ADR-0028). 19.0 glazed_type codes (1-4,6,7) are a subset of the verified 1-8
space. The 6 rich certs use lodged window_area directly; the windowless 994
synthesise a 4-way N/E/S/W split.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Dispatch RdSAP-Schema-19.0 through from_api_response, parse-fix the schema
(data-driven required->optional, validated against the 1000-cert 19.0 corpus
per ADR-0028), and port 18.0's defensive mapper reads (dwelling_type str/dict/
number, photovoltaic_supply guard, sap_room_in_roof Measurement coercion).
All 1000 corpus certs now parse and map.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Orchestrator runs recommend_secondary_heating_removal; report._triggers_for
explains it via the lodged secondary_heating_type; harness catalogue + ARA seed
price it. Re-pins the golden/integration plans it shifts: it is a cheap (\£250)
SAP lever, so on gas-main certs lodging an electric secondary (691) it displaces
the \£12k ASHP (0330, 0036) or joins the all-beneficial-measures package (000490,
where its marginal SAP is 0 under the category-4 ASHP but the heater is still
physically removed). Consistent with the optimiser's existing kitchen-sink
package behaviour.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two cascade tests on the worksheet-pinned 001431 build_epc() (the user's
before/after Summary PDFs trip the documented 001431 window-extraction bug, so
the repo's sanctioned 001431 baseline is used instead):
- electric-storage main (code 402) + secondary 691: removal reproduces the
secondary-removed cert at delta 0 — RdSAP §A.2.2 re-forces a default secondary,
matching the user's F35→F35 example;
- gas combi main (code 104) + secondary 691: removal strictly raises SAP
(74.22→77.61) — the Table 11 fraction reallocates to the cheaper main.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Records the four load-bearing design decisions from the grill-with-docs session:
standalone co-selectable rec; eligibility = lodged-only (no effectiveness gate,
electric-storage §A.2.2 no-op is the Optimiser's call); dedicated clearing
SecondaryHeatingOverlay; flat per-dwelling cost (a lodged secondary is fixed per
RdSAP, so a real decommission job, not room-scaled).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Flat per-dwelling decommission price (sample_catalogue \£250) + 0.25 contingency
(covers unknown heater count / hard-wired-vs-plugged / repaint extent). The JSON
repo joins the contingency from config, proven by the new repo test. No composite
Products machinery — a lodged secondary is one roughly-fixed job, not room-scaled.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
recommend_secondary_heating_removal offers one standalone Option that clears the
lodged secondary system. Eligibility is purely physical (offer iff
sap_heating.secondary_heating_type is set) — no effectiveness gate, since a
lodged secondary is a fixed emitter per RdSAP (portables are ignored), and the
electric-storage §A.2.2 no-op is the Optimiser's call (ADR-0028 decisions 1-2).
Priced at a flat per-dwelling decommission cost, not room-scaled.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
ADR-0028's deferred extraction, triggered by 17.1 as the second instance: the
inherited 20.0.0 coefficients (0.148 + band multipliers + 4-way split) now live
in one `_synthesise_reduced_field_windows` core. The 20.0.0 / 18.0 / 17.1 seams
keep their own names (so each can diverge) but collapse to glazing-type
resolution (cascade + that spec's ND handling) plus a call to the core.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
