Gold analytics: anomalies, correlations, sleep, training, supplements, SLA

[EduardPetraeus]

Context (updated 2026-03-05)

With AI-first architecture (ADR-005), analytics are split:

• Local: AI agent computes on-demand via semantic contracts + MCP tools. No materialized views needed.
• Cloud: Databricks Gold views for BI consumers.

Remaining Scope (Databricks only)

• gold.daily_anomalies — z-score per metric
• gold.metric_correlations — cross-metric Pearson r
• gold.sleep_quality_analysis — sleep architecture breakdown
• gold.training_recovery_balance — ATL/CTL/TSB
• gold.supplement_effectiveness — supplement vs biomarker correlation
• gold.data_sla_status — freshness monitoring per source

Local Equivalent

• discover_correlations MCP tool handles correlation analysis
• Anomaly detection via personal baselines in ai/baseline_computer.py
• Sleep and supplement analysis via semantic contracts

[EduardPetraeus]

Done in Session 3: 7 fact views + anomaly/correlation/sleep analytics built into gold_local/. Part of Gold Local Views.