Household energy forecaster

Enter the last hour of whole-house power readings. The model forecasts the next 30 minutes for the whole house and for 24 appliance channels. Everything runs in this browser tab, so your data is never uploaded.

Starting…

Input

Total household power for the last 12 readings, in watts, oldest first. Each reading is a 5-minute average. One per line or comma-separated. Time of the most recent reading, as the clock shows it in your data. No timezone conversion is applied.

How to create scalers.json

The model was trained on standardised data, so the page needs the same scaler values. Run this in your training notebook (after the scalers are fitted) and put the file next to this page.

import json
json.dump({
    "x_mean":  scaler_X.mean_.tolist(),
    "x_scale": scaler_X.scale_.tolist(),
    "y_mean":  scaler_Y.mean_.tolist(),
    "y_scale": scaler_Y.scale_.tolist(),
}, open("scalers.json", "w"))

Forecast

Channel

Run a forecast to see it here.

Input

Forecast

All channels