DigiNews

Tech Watch by Johan Denoyer

← Back to articles

Predicting home electricity usage based on historical patterns in Home Assistant

Quality: 6/10 Relevance: 8/10

Summary

This post outlines a half-hourly load predictor for Home Assistant that uses historical patterns by day of week and time slot. It blends recent data (LAST 7 days) with older data (7-56 days) using a weighted scheme and a SQL implementation to produce a lookup of predicted kWh per half-hour slot. It’s a practical example of time-series heuristics and data engineering in home automation, suitable for DIY energy optimization but not a formal AI model.

🚀 Service construit par Johan Denoyer