Rules

Formula: ZScore

What it does:
This rule uses statistical analysis to detect anomalies based on how far the latest value deviates from the average usage, measured in standard deviations (called the Z-score). If the Z-score’s absolute value exceeds your selected threshold, an alert is triggered.

When to use it:
Use ZScore when you want to detect outliers in a more intelligent way—ideal for identifying rare spikes or drops that are unusual given your usage patterns.

Examples

✅ Example 1: Large Increase Triggers Alert

• Anomaly Formula: ZScore
• Change Type: Increased
• Threshold: 2
• History: 50, 52, 49, 51, 48, 100
• Explanation:
  • Mean ≈ 50
  • Standard deviation ≈ 17.7
  • Latest = 100
  • Z-score ≈ (100 – 50) / 17.7 ≈ 2.82
  • Above threshold and it’s an increase → ✅ Alert triggered

❌ Example 2: Drop Too Small to Trigger Alert

• Anomaly Formula: ZScore
• Change Type: Decreased
• Threshold: 3
• History: 60, 62, 58, 61, 59, 56
• Explanation:
  • Mean ≈ 59.3
  • Standard deviation ≈ 2.0
  • Latest = 56
  • Z-score ≈ (56 – 59.3) / 2.0 ≈ -1.65
  • Drop is within expected range → ❌ No alert

✅ Example 3: Sharp Outlier Detected (Change Type = Any)

• Anomaly Formula: ZScore
• Change Type: Any
• Threshold: 2.5
• History: 70, 72, 71, 68, 73, 20
• Explanation:
  • Mean ≈ 70.6
  • Standard deviation ≈ 4.8
  • Latest = 20
  • Z-score ≈ (20 – 70.6) / 4.8 ≈ -10.54
  • Extremely unusual drop and change type is Any → ✅ Alert triggered