Scoring and Keys

This page documents the scoring behavior and emitted metric keys for ml_architecture, aligned to:

crates/arxo-engine/src/metrics/ai_observability/ml_architecture/plugin.rs
crates/arxo-engine/src/metrics/ai_observability/ml_architecture/ui/health_summary.rs

Overall Score Formulas

ml_architecture.overall_score is confidence-weighted and clamped to 0..1.

For each detector score s_i with detector confidence c_i:

For each detector group G:

Overall:

overall_score = clamp(
  0.20 * group1 +
  0.35 * group2 +
  0.45 * group3,
  0.0,
  1.0
)

Group membership:

group1 (0.20): train_serve_skew_risk, skew_test_absence, train_inference_boundary
group2 (0.35): pipeline_complexity, reproducibility, data_lineage_integrity, experiment_isolation, eval_integrity
group3 (0.45): serving_maturity, drift_monitoring, data_validation, ci_integration, monitoring_alerting, model_staleness, serving_ops, shadow_canary, ab_testing, fairness_audit

ml_architecture.overall_score_extended keeps the same base groups and adds two additive groups:

group4 (0.20): model_validation_gates, calibration_uncertainty, feature_store_consistency, progressive_delivery_analysis, provenance_attestation, responsible_ai_governance
group5 (0.20): attestation_enforcement, model_registry_governance, lineage_schema_fidelity, adversarial_resilience, post_market_incident_readiness, genai_telemetry_semconv

If no training/serving ML files are detected:

Metric Key
`ml_architecture.train_serve_skew_score`
`ml_architecture.skew_test_absence_score`
`ml_architecture.pipeline_complexity_score`
`ml_architecture.reproducibility_score`
`ml_architecture.train_inference_boundary_score`
`ml_architecture.data_lineage_integrity_score`
`ml_architecture.experiment_isolation_score`
`ml_architecture.eval_integrity_score`
`ml_architecture.serving_maturity_score`
`ml_architecture.drift_monitoring_score`
`ml_architecture.data_validation_score`
`ml_architecture.ci_integration_score`
`ml_architecture.fairness_audit_score`
`ml_architecture.ab_testing_score`
`ml_architecture.shadow_canary_score`
`ml_architecture.monitoring_alerting_score`
`ml_architecture.model_staleness_score`
`ml_architecture.serving_ops_score`
`ml_architecture.model_validation_gates_score`
`ml_architecture.calibration_uncertainty_score`
`ml_architecture.feature_store_consistency_score`
`ml_architecture.progressive_delivery_analysis_score`
`ml_architecture.provenance_attestation_score`
`ml_architecture.responsible_ai_governance_score`
`ml_architecture.attestation_enforcement_score`
`ml_architecture.model_registry_governance_score`
`ml_architecture.lineage_schema_fidelity_score`
`ml_architecture.adversarial_resilience_score`
`ml_architecture.post_market_incident_readiness_score`
`ml_architecture.genai_telemetry_semconv_score`

Metric Key	Range / Type	Direction
`ml_architecture.overall_score`	`0..1`	Higher is better
`ml_architecture.overall_score_extended`	`0..1`	Higher is better
`ml_architecture.gpu_file_count`	Number	Informational
`ml_architecture.database_file_count`	Number	Informational
`ml_architecture.env_config_file_count`	Number	Informational
`ml_architecture.graph.*`	Graph entries	Informational

This contract is documented against metric version 2.0.0.