Green Home Score
Submission D, E and G

Methodology & evaluation

Read live from the running engine.

Green Home Score weights

S = Σ wₖsₖ, weights add up to 100%

Sub-score mappings

A better state never scores lower (property-tested)

Held-out test metrics vs simple baselines

Property-level splits, so no home appears in two

AI assistant

off, template fallback
  1. Engines calculate. Score, savings, carbon and value all come from the models above.
  2. The AI explains. It only writes about numbers already in the report.
  3. Every number is checked. One that does not match gets one retry, then a fixed template is used.
  4. What-ifs stay grounded. The AI only picks upgrades from a fixed list and the engines recalculate.

Emission factors

Database tables, created on first use

Checking…