Jira progress: loading…
RCAS
Root Cause Analysis Engines
1. Overview
Root Cause Analysis Engines (RCAS) are assistive calculation engines that analyze signals, execution outcomes, and metadata to diagnose anomalies and propose likely root causes.
RCAS engines support investigation and remediation but do not modify data directly.
2. Design Principles
-
Diagnostic Focus
RCAS engines explain why something may have occurred. -
Hypothesis-Driven
Outputs represent plausible causes, not guaranteed truth. -
Cross-Signal Reasoning
Analysis may span multiple signals, engines, and layers.
3. Scope of Responsibility
3.1. What RCAS Engines Do
- Analyze anomalies or unexpected results
- Correlate issues across signals and pipelines
- Propose likely root causes or contributing factors
- Suggest corrective actions or follow-up checks
Typical use cases:
- Investigating data quality issues
- Explaining sudden metric changes
- Supporting assurance and audit workflows
4. What RCAS Engines Do Not Do
- ❌ Modify or recompute data
- ❌ Enforce corrective actions
- ❌ Emit authoritative signals
- ❌ Replace human judgment
5. Inputs
RCAS engines consume:
- Signal values and metadata
- Execution logs and provenance
- Historical baselines and thresholds
6. Outputs
RCAS engines produce:
- Root cause hypotheses
- Supporting evidence references
- Confidence scores and explanations
Outputs are surfaced via:
- ZARA explainability views
- Verifier or operator workflows
7. Governance & Trust
RCAS outputs are:
- explicitly marked as diagnostic,
- confidence-scored,
- intended for human review.
They must never silently alter authoritative results.
8. Canonical Identification
- Engine Type:
RCAS - USO Code:
RCAS - Category: Assistive Calculation Engine
- Layer: Computation Hub
9. Related Engine Types
- AGGR — Aggregation Engines
- CFIL — Confidence Filter Engines
- ZARA — Orchestration & Explainability
Status: Stable
Owner: Intelligence Governance / ZARA