The asf_heat_pump_suitability repo contains the code used to calculate heat pump suitability scores for lower-layer
super output areas (LSOAs) in England and Wales using domestic EPC data and supplementary sources. Scores are
weight-adjusted for LSOAs where possible to reduce bias.
Read more about the project here.
- Meet the data science cookiecutter requirements, in brief:
- Install:
direnvandconda
- Install:
- Clone the repo and navigate to your local repo folder
- Run
direnv allow - Run
make installto configure the development environment:- Setup the conda environment
- Configure
pre-commit - Install requirements
- Run
conda activate asf_heat_pump_suitability - Instructions to run pipeline scripts can be found in asf_heat_pump_suitability/pipeline/README.md
See the general repository structure depicted below. Key files are also shown.
asf_heat_pump_suitability
├───analysis/
│ Scripts for ad-hoc analysis
├───config/
│ Respository config files and global variables
│ ├─ base.yaml - data sources and mappings
│ ├─ README.md - data source information, citations, and attributions
├───getters/
│ Modules with functions to load data
│ ├─ base_getters.py - generic getter functions
│ ├─ get_datasets.py - specific getter functions to load raw datasets
│ ├─ get_target.py - specific getter functions to load and process target data for reweighting
├───notebooks/
│ Notebooks with prototype code for pipeline
├───pipeline/
│ Subdirs with modules to process data and produce outputs
│ ├─ evaluation/
│ ├─ prepare_features/
│ ├─ reweight_epc/
│ ├─ run_scripts/
│ ├─ sampling/
│ ├─ suitability/
│ ├─ README.md - instructions to run pipeline
├───utils/
│ Modules with generic utils
One of the challenges in assessing heat pump suitability is to set criteria for what makes a home suitable for a particular technology. We have used two sets of criteria in this project: one a “conventional” view, which we think reflects common consensus; and one a Nesta view, which draws on our latest research. We did this for four different technologies; air source heat pumps (ASHPs), ground source heat pumps (GSHPs), heat networks (HNs) and shared ground loops (SGLs).
This pipeline therefore computes a conventional score and a Nesta score for each of the four tech types listed: eight heat pump suitability scores are calculated in total per LSOA. Scores are first computed per property based on presence/ absence of certain characteristics of the property/area using a simple additive model (see table below). Scores are then averaged per property before finally aggregating to LSOA level.
| ASHP (S) | ASHP (N) | GSHP (S) | GSHP (N) | SGL (S) | SGL (N) | HN (S) | HN (N) | |
|---|---|---|---|---|---|---|---|---|
| Is the property NOT listed? | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 |
| Is the property NOT in a building conservation zone? | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 | 0.25 |
| Is the EPC rating >= C? | 1 | 0 | 1 | 0 | 1 | 0 | 0 | 0 |
| Is the property NOT a flat? | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 |
| Is the garden >10m2? | 1 | 0 | 1 | 0 | 1 | 0 | 0 | 0 |
| Is there >2m2 of external space? | 0 | 2 | 0 | 1 | 0 | 0 | 0 | 0 |
| Is it off-gas? | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 |
| Is this property part of a building with multiple other properties (e.g. a flat)? | 0 | 0 | 0 | 0 | 2 | 2 | 2 | 2 |
| Is there a high property density (>60 households per km2) in this LSOA? | 0 | 0 | 0 | 0 | 2 | 2 | 0 | 0 |
| Is this property in an urban LSOA/high heat demand density LSOA? | 0 | 0 | 0 | 0 | 0 | 0 | 2 | 2 |
| Maximum points per property | 4 | 4 | 4 | 3 | 7 | 5 | 5 | 5 |
A comprehensive table of citations for data used in this analysis can be found in asf_heat_pump_suitability/config/README.md. See attributions below.
- Contains OS data © Crown copyright and database right 2024.
- Contains Royal Mail data © Royal Mail copyright and database right 2024.
- Contains Office for National Statistics information licensed under the Open Government Licence v.3.0.
- Contains public sector information licensed under the Open Government Licence v3.0.
- Contains GeoPlace data © Local Government Information House Limited copyright and database right 2024.
- This work uses HM Land Registry's INSPIRE Index Polygons service. This information is subject to Crown copyright and database rights 2024 and is reproduced with the permission of HM Land Registry. The polygons (including the associated geometry, namely x, y co-ordinates) are subject to Crown copyright and database rights 2024 Ordnance Survey 100026316. INSPIRE Index Polygons service Conditions of use.
- Microsoft GlobalMLBuildingFootprints are made available under the Open Database License. Any rights in individual contents of the database are licensed under the Database Contents License: https://opendatacommons.org/licenses/dbcl/1.0/\ .
This work uses Facebook Research's balance package and ipfn to conduct iterative proportional fitting.
Sarig, T., Galili, T., & Eilat, R. (2023). balance – a Python package for balancing biased data samples. https://arxiv.org/abs/2307.06024
Intermediate outputs include:
- EPC data weighted according to LSOA using Iterative Proportional Fitting to reduce bias
- EPC data enhanced with new features including: lat/lon; listed building status; building conservation zone status; off gas status; average garden size per MSOA; property density per LSOA
- Individual garden size estimates for UPRNs in EPC
Technical and working style guidelines
Project based on Nesta's data science project template (Read the docs here).