Infrastructure · Data Structures
Anomaly
Datum whose observed value deviates from the expected baseline, triggering investigation
Definition
Mechanism
A Datum whose observed value falls outside the expected baseline by a measurable deviation score. Anomalies carry the (observed, expected, deviation) triple as first-class structure, so they can be routed, ranked, and triaged. Serves as the trigger artifact for investigation, learning, and diagnostic workflows.
Invariants
None stated.
Risks
Failure modes
None stated.
Relationships
Builds on
References
Raw unit of fact
Used by
References
Continuous observation of state over time
Mind
Restructuring categories to fit data
Reference
Data schema
{
"type": "object",
"required": [
"observed_value",
"expected_baseline",
"deviation_score"
],
"properties": {
"observed_value": {
"description": "The data point that triggered detection (any type)"
},
"expected_baseline": {
"description": "The prediction or mean value (any type)"
},
"deviation_score": {
"type": "number",
"description": "Standard deviations (sigma) or distance metric"
},
"detected_at": {
"type": "string",
"format": "date-time"
}
}
}