Rules

Anomaly Formula: Modified-ZScore

What it does:
This rule detects outliers using the Modified Z-score, which is based on the median and MAD (Median Absolute Deviation) instead of the mean and standard deviation. This makes it more robust against extreme values in your data.

When to use it:
Use this when you want to catch extreme deviations while minimizing the effect of noisy or highly skewed data. It’s especially useful if you expect your usage data might include sharp, one-off spikes.

Examples

✅ Example 1: Strong Increase Detected

• Anomaly Formula: ModifiedZScore
• Change Type: Increased
• Threshold: 3
• History: 50, 52, 51, 48, 49, 100
• Explanation:
  • Median ≈ 50.5
  • MAD ≈ 1.5
  • Latest = 100
  • Modified Z-score ≈ (0.6745 * (100 – 50.5)) / 1.5 ≈ 22.4
  • Z-score is above threshold → ✅ Alert triggered

❌ Example 2: Drop But Change Type Is ‘Increased’ (No Alert)

• Anomaly Formula: ModifiedZScore
• Change Type: Increased
• Threshold: 3
• History: 60, 59, 61, 58, 62, 40
• Explanation:
  • Latest = 40, which is a drop
  • Change type is set to “Increased” → ❌ No alert triggered

✅ Example 3: Unusual Drop Caught with Change Type = Any

• Anomaly Formula: ModifiedZScore
• Change Type: Any
• Threshold: 3
• History: 75, 78, 76, 80, 77, 20
• Explanation:
  • Median ≈ 77
  • MAD ≈ 2
  • Latest = 20
  • Modified Z-score ≈ (0.6745 * (20 – 77)) / 2 ≈ -19.57
  • Absolute Z-score is well above threshold → ✅ Alert triggered