Help & FAQ · EcoEdge

01 · Overview

What is EcoEdge?

EcoEdge learns how your home holds heat and recommends the lowest setpoint that keeps you comfortable. It reads your thermostats through the Home Assistant integration, fits a physics model of each room, and shows where energy is being spent that doesn't need to be.

Recommendations are advisory — EcoEdge never changes your thermostats. No manual configuration is needed: the models learn automatically from your data.

02 · The setpoint chart

What are the lines?

Your setpoint (solid)

The temperature your thermostat is currently set to in Home Assistant. This controls your heating right now.

AI recommendation (dashed)

What EcoEdge recommends based on your room's thermal model and current weather. If it sits below your setpoint, your home may be heating more than needed.

Forecast (dotted, after the NOW marker)

Predicted indoor temperature and the hour-by-hour recommendation for the next 24 hours, driven by the weather forecast and your home's thermal profile.

The gap between your setpoint and the AI recommendation is where energy savings come from.

03 · Heat loss

The k value

Each room gets a heat-loss coefficient k, fitted from how fast the room cools when heating is off relative to the indoor–outdoor temperature difference:

dT/dt ≈ −k · (T_indoor − T_outdoor)

The higher the k, the more energy your home needs to stay warm.

Band	k value	Meaning
Well insulated	k < 0.3	Holds heat — the AI can relax setpoints further.
Average	0.3 – 0.6	Typical envelope — moderate savings headroom.
High loss	k > 0.6	Loses heat fast — recommendations stay conservative.

04 · Learning models

How EcoEdge learns your home

Two physics models are fitted continuously against your history, and a machine-learning model observes across homes. The best-fitting physics model drives the recommendation.

Model	What it does	Role
RC	Newton cooling — measures heat loss from idle periods.	Physics
KQ	RC plus a heating-power term — learns from mixed heating/idle periods.	Physics
ML	Cross-home learning that predicts heat loss from weather and building behavior.	Observing

You always see one saving estimate. When the ML layer is active, an ML tag appears — it means the recommendation is enhanced by cross-home learning on top of your room's physics model.

05 · Connecting a home

Install the integration

Open HACS in Home Assistant → Integrations → custom repositories → add https://github.com/ecoedge-dev/ha_ecoedge_ai_thermostat (category Integration).
Download EcoEdge AI Thermostat and restart Home Assistant.
Settings → Devices & Services → Add Integration → search "EcoEdge AI Thermostat".
Sign in with the same email and password you use here.
Select the thermostats to optimise and an outdoor temperature sensor, then submit.
Your home appears on the dashboard once the first data batch arrives.

The integration also creates sensor entities in Home Assistant (AI setpoint, model, heat loss, confidence, saving estimate) that you can use in your own automations.