bf99b1c7df
The 7-cert ASHP+battery PE cluster was overshooting by +2.7..+8.1 kWh/m²
after the PE β-split landed in S0380.45. The handover hypothesised an
E_PV magnitude bug ("cascade thinks 2570 kWh/yr vs worksheet 831"). The
worksheet PDF for cert 0380 (dr87-0001-000899.pdf line 233) was
verified to show **-2563.3692** kWh/yr — matching our cascade. The
real bug was different: the **5-kWh battery wasn't reaching the
cascade**, so β-coefficients used the no-battery branch (C1=1.61,
β≈0.36) instead of the 5-kWh branch (C1=1.12, β≈0.75).
Per SAP 10.2 Appendix M1 §3c-d (p.94): "C_bat is the usable capacity
of the battery in kWh, limited to a maximum value of 15 kWh. C_bat=0
if no battery present." Cert 0380 lodges `pv_battery_count: 1` and
`pv_batteries: [{"battery_capacity": 5}]` — but the schema's
`PvBatteries` dataclass had only `pv_battery: Optional[PvBattery]`,
matching the older synthetic fixture shape (nested
`{"pv_battery": {"battery_capacity": 5}}`). The real-API payload's
flat `battery_capacity: 5` was silently dropped during `from_dict`.
Two surgical changes:
- `datatypes/epc/schema/rdsap_schema_21_0_1.py`: add
`battery_capacity: Optional[float] = None` as a sibling to
`pv_battery` on `PvBatteries`. Synthetic-shape certs continue to
populate the nested form; real-API certs now populate the flat form.
- `datatypes/epc/domain/mapper.py:_first_pv_battery`: prefer nested
when present, fall back to the flat lifted field. Domain still
exposes a single uniform `PvBatteries(pv_battery=PvBattery(...))`
shape downstream.
Cohort impact (PE residual kWh/m² vs worksheet):
| Cert | Pre-S0380.48 | Post-S0380.48 |
|---|---:|---:|
| 0350 | +2.73 | -3.58 |
| 0380 | +8.09 | -4.01 |
| 2225 | +4.48 | -4.50 |
| 2636 | +3.42 | -4.14 |
| 3800 | +3.58 | -4.01 |
| 9285 | +3.20 | -3.46 |
| 9418 | +4.67 | -3.76 |
Cluster magnitude dropped from +2.7..+8.1 to -3.5..-4.5 — the cascade
now over-credits PV by ~4 kWh/m² (vs previously under-crediting by
~5 kWh/m²). The residual flipped sign because cascade β=0.75-0.81
slightly exceeds worksheet β=0.74 (read from page-3 line 233a/233b
ratio 1903.39/2563.37 = 0.7426). The remaining ~4 kWh/m² under-shoot
traces to two structural factors deferred until a fresh closure
slice ships:
1. The synthetic-default `pv_export_primary_factor = 0.501` is the
annual Table 12 code-60 value. The worksheet uses the effective
monthly Table 12e factor weighted by E_PV,ex,m (cert 0380: 0.4268
= -0.074 differential). The cascade's `_effective_monthly_pe_
factor` already computes the same weighting for PV — but the
calculator's PV PE credit reads `inputs.other_primary_factor`
(=1.501) and `inputs.pv_export_primary_factor` (=0.501) directly,
bypassing the per-end-use effective-monthly cascade.
2. Cascade β slightly higher than worksheet (0.751 vs 0.7426 on
cert 0380) — likely a monthly-distribution detail in D_PV.
SAP scores remain exact across the cohort (residual +0 every cert).
CO2 residuals all <0.11 t/yr (well within the 0.001-tolerance pin
range after re-pin). 9501 (PV no battery) preserved at +0.255 PE /
-0.047 CO2 — no regression. Re-pins all 7 golden fixtures in the
same slice per [[feedback-commit-per-slice]].
Pyright net-zero on touched files (32 errors before, 32 after).
|
||
|---|---|---|
| .devcontainer | ||
| .github/workflows | ||
| .idea | ||
| .vscode | ||
| applications | ||
| asset_list | ||
| backend | ||
| backlog | ||
| datatypes | ||
| deployment/terraform | ||
| docs/adr | ||
| domain | ||
| epr_data_exports | ||
| etl | ||
| infrastructure | ||
| model_data/requirements | ||
| orchestration | ||
| recommendations | ||
| repositories | ||
| scripts | ||
| sfr/principal_pitch | ||
| survey_report | ||
| tests | ||
| utilities | ||
| utils | ||
| .coveragerc | ||
| .dockerignore | ||
| .gitignore | ||
| __init__.py | ||
| ara_backend_design.md | ||
| BaseUtility.py | ||
| CLAUDE.md | ||
| conftest.py | ||
| CONTEXT.md | ||
| devcontainer.sh | ||
| Dockerfile.test | ||
| Dockerfile.test.dockerignore | ||
| Makefile | ||
| MEMORY.md | ||
| package-lock.json | ||
| package.json | ||
| pyproject.toml | ||
| pyrightconfig.json | ||
| pytest.ini | ||
| README.md | ||
| run_lambda_local.sh | ||
| serverless.yml | ||
| test.requirements.txt | ||
| tox.ini | ||
| UBIQUITOUS_LANGUAGE.md | ||
Model Repository
This repository contains the code pertaining to the development of the data science and machine learning products being utilised by Hestia.
The different folders in this repository relate to services that can be used independently, or can be imported and used as part of a larger application
Getting Started
Prerequisites
Dev Container Setup
This repo uses a Docker Compose-based dev container. The model-backend service joins a shared-dev Docker network so it can communicate with other local services (e.g. a frontend container) running on your machine.
VS Code users: The initializeCommand in devcontainer.json creates the shared-dev network automatically before the container starts. No manual step required — just open the repo and select Reopen in Container.
Non-VS Code / CI workflows: Run the following once before starting the container:
make dev-setup
This is idempotent and safe to re-run if the network already exists.
Folders
backend/
This folder contains the code for the fastapi backend service, which provides an interface to much of the functionality in this repository, for the frontend
model_data/
This folder contains related to the reading and preparation of assessment model data, including pulling out epc attributes
Testing
All tests can be run, against the configuration in pytest.ini running
pytest
This will run the complete panel of tests and report on coverage in the locations specified by the pytest.ini file.
To run tests in a specific service, e.g. inside of model_data, simply run
pytest --cov-config=model_data/.coveragerc --cov=model_data
This will produce the test results and coverage reports