How do seasonal adjustments affect my alarm thresholds?
Seasonal adjustments modify your alarm's expected baseline to account for natural consumption changes throughout the year. Without them, a heating alarm set for winter levels would trigger constantly in summer (when heating is low) or miss genuine spikes in winter (when heating is naturally high).
How seasonal factors work
The system divides the year into four seasons, each with a multiplier:
- Winter (December–February): e.g., 1.3 means 30% above annual average
- Spring (March–May): e.g., 1.0 means at average
- Summer (June–August): e.g., 0.6 means 40% below average
- Fall (September–November): e.g., 0.9 means 10% below average
The multiplier is applied to the baseline (pattern mean or rolling average) before comparing against the actual reading. So if your baseline is 100 kWh and the winter factor is 1.3, the expected value becomes 130 kWh.
Where seasonal factors come from
- Pattern alarms: Seasonal factors are calculated automatically during pattern learning. The model analyzes your 8 weeks of historical data and computes the ratio of each season's average to the overall average.
- Trend alarms: You enable seasonal adjustment with the use_seasonal_adjustment toggle. The system applies learned factors or defaults to 1.0 if no pattern model exists.
When to enable seasonal adjustment
- Enable for heating, cooling, and gas meters — these have strong seasonal patterns
- Enable for buildings with seasonal occupancy (schools, holiday properties)
- Disable for electricity in always-occupied buildings — electricity usage is often more consistent year-round
- Disable for water meters unless you have irrigation or seasonal outdoor use
Example: Your heating meter averages 200 kWh. With a 20% deviation threshold and winter factor 1.3:
- Without seasonal adjustment: threshold = 200 × 1.2 = 240 kWh (would trigger constantly in winter when normal heating is 260 kWh)
- With seasonal adjustment: threshold = 200 × 1.3 × 1.2 = 312 kWh (correctly accounts for higher winter usage)
Where to configure: Analytics → Alarms → Trend-Based tab → enable "Seasonal Adjustment" toggle.