Human behavior is composed of Habit Loops: Cue (trigger signal) → Routine (behavioral pattern) → Reward (outcome). Smart homes that learn Habit Loops can anticipate and manage energy proactively — with far greater precision than systems that merely detect motion.
Three Levels of Habit Loops IoT Learns
Level 1: Daily Habit Loops — Routines that repeat every day: always turning on the bathroom air conditioner and kettle simultaneously upon waking; watching television daily from 19:30–21:00; turning off bedroom lights at the alarm sound. The system learns these patterns within 2–3 weeks and manages energy automatically.
Level 2: Weekly Habit Loops — Routines that differ by day of the week. Monday–Friday the occupant leaves at 08:00; Saturday they may sleep until 09:30. The system tailors Pre-Cool schedules to each specific day rather than applying one universal schedule — saving an additional 8–12% HVAC energy by separating Weekday and Weekend profiles.
Level 3: Contextual Habit Loops — Routines that change based on external context. When Bangkok AQI exceeds 100, occupants tend to stay home and run air purifiers longer than usual. When it rains, HVAC load decreases because outdoor temperatures drop. AI learns these correlations and adjusts energy management based on actual environmental conditions.
Semi-Supervised Learning for Edge Cases
When the system encounters unfamiliar behavior — such as guests staying for several days — Semi-Supervised Learning activates a temporary Learning Mode, capturing the exceptional period data separately without corrupting the normal Baseline Profile, then resuming standard profiles when the exceptional period ends.
Combined Outcome of Habit Loop Learning
Adding the Habit Loop Layer on top of foundational Behavioral Learning increases overall energy efficiency by an additional 10–18% with no additional commands from occupants. The system generates a weekly Habit Insight report via LINE OA informing homeowners what new patterns were learned.