fusion-bicc-data-drift-detect

## Copyright (c) 2026, Oracle and/or its affiliates. ## Licensed under the Universal Permissive License v 1.0 as shown at http://oss.oracle.com/licenses/upl # Fusion BICC Data Drift Detection Use this skill when schema appears stable but data behavior changes across extraction windows and can impact KPIs, model quality, or ODT load quality. ## When to use - Null rates spike after a Fusion release. - Distinct cardinality unexpectedly collapses or expands. - Measure ranges/means shift materially. - You need severity-rated drift findings with remediation actions. ## Inputs - Baseline and current BICC CSV extracts. - Data drift policy (`assets/data_drift_policy.yaml`). ## Workflow 1. Build baseline data profile snapshot from known-good extraction window. 2. Build current data profile snapshot for the latest run. 3. Run drift detection with policy thresholds. 4. Review critical findings and associated remediation in report outputs. ## Commands ```bash # 1) Build profile snapshots python skills/fusion-bicc-data-drift-detect/scripts/build_data_profile_snapshot.py \ --input-dir ./inputs/baseline \ --output ./outputs/baseline_data_profile.json python skills/fusion-bicc-data-drift-detect/scripts/build_data_profile_snapshot.py \ --input-dir ./inputs/current \ --output ./outputs/current_data_profile.json # 2) Detect data drift python skills/fusion-bicc-data-drift-detect/scripts/detect_data_drift.py \ --baseline ./outputs/baseline_data_profile.json \ --current ./outputs/current_data_profile.json \ --policy skills/fusion-bicc-data-drift-detect/assets/data_drift_policy.yaml \ --output-json ./outputs/data_drift_report.json \ --output-md ./outputs/data_drift_report.md ``` ## Outputs - `data_drift_report.json` - `data_drift_report.md` ## Critical examples - Required field null ratio exceeds critical threshold. - Row count drops below expected freshness/completeness threshold. - Categorical domain shifts beyond policy tolerance. See `references/data-drift-metrics.md` and `references/severity-model.md`.