How do trend-based alarms compare consumption to historical baselines?

Trend-based alarms detect when your current consumption deviates significantly from historical patterns. Instead of a fixed threshold, they use a calculated baseline from your past data.

Two baseline methods

1. Year-over-Year (YoY): Compares current readings against the same period last year. Best for seasonal buildings where consumption naturally varies by season (e.g., heating peaks in winter). Requires at least 12 months of data.

2. Rolling Average: Compares against an average of recent weeks. You choose the period — 4, 8, 12, 24, or 52 weeks. Best for detecting gradual drift or sudden spikes against recent norms. A 12-week rolling average is the default.

Configuration options (trend_config)

• baseline_method: "yoy" or "rolling_avg"
• rolling_period_weeks: Number of weeks for rolling average (4–52)
• deviation_percent: How far current reading must deviate to trigger (e.g., 20% means a reading 20% above baseline triggers the alarm)
• comparison_direction: "above" (only high usage), "below" (only low usage — useful for detecting meter failures), or "both"
• use_seasonal_adjustment: Apply seasonal multipliers (winter, spring, summer, fall) to the baseline for more accurate comparison

How to set up a trend alarm

1. Go to Analytics → Alarms → click the "Trend-Based" tab
2. Select your meter from the dropdown
3. Choose a time range for visualization (1 month, 3 months, 6 months, or 1 year)
4. Configure the baseline method, deviation percentage, and comparison direction in the left panel
5. Click "Preview Changes" to see the trend line and threshold bands on the chart
6. Click "Create & Enable Alarm" to activate

The chart shows: Your actual consumption as data points, the calculated baseline as a line, and upper/lower threshold bands. Points outside the bands would trigger alarms.

Where to manage: Analytics → Alarms → Trend-Based tab.