Khalim Conn-Kowlessar 027ee1fba3 refactor(epc-prediction): extract shared leave-one-out scorer + corpus loader (ADR-0030) ... "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>		2026-06-14 09:12:08 +00:00
..
__init__.py	feat(epc-prediction): Comparable Properties selection ladder (ADR-0029)	2026-06-13 23:44:57 +00:00
test_comparable_properties.py	fix(epc-prediction): dedupe re-lodgements + leak-free leave-one-out (ADR-0029)	2026-06-14 00:40:23 +00:00
test_epc_prediction.py	feat(epc-prediction): cohort-mode the roof/floor/insulation/age categoricals (ADR-0029)	2026-06-14 00:31:16 +00:00
test_prediction_comparison.py	feat(epc-prediction): complete component coverage — fabric/glazing/renewables/doors (ADR-0030)	2026-06-14 09:00:30 +00:00
test_validation.py	refactor(epc-prediction): extract shared leave-one-out scorer + corpus loader (ADR-0030)	2026-06-14 09:12:08 +00:00