Overview — 전체 아키텍처 소개

# 노트북 목차 — 각 단계에서 사용하는 Databricks 기능
notebooks = {
    "02_feature_engineering":   ["Delta Lake", "Feature Store", "Unity Catalog"],
    "03_model_training":        ["MLflow Tracking", "Autolog", "Data Lineage"],
    "03a_ml_trends":            ["ML 최신 기술 트렌드", "알고리즘 진화 역사"],
    "03b_multi_algorithm":      ["MLflow 실험 비교 UI"],
    "03c_advanced_techniques":  ["Databricks AutoML", "Optuna", "SMOTE"],
    "03d_retraining_strategies":["재학습 전략", "Incremental/Continual/Online Learning"],
    "04_model_registration":    ["UC Model Registry", "Alias", "Lineage"],
    "05_challenger_validation": ["mlflow.evaluate()", "태그 기반 검증"],
    "06_batch_inference":       ["PySpark UDF", "Delta Lake"],
    "07_anomaly_detection":     ["Volumes", "GPU Cluster", "MLflow"],
    "08_model_monitoring":      ["Lakehouse Monitoring", "PSI"],
    "09_mlops_agent":           ["AI Agent", "UC Functions as Tools"],
    "10_job_scheduling":        ["Databricks Workflows"],
}

-- 고장 유형별 분포 확인
SELECT
  machine_failure,
  SUM(twf) as tool_wear_failure,
  SUM(hdf) as heat_dissipation_failure,
  SUM(pwf) as power_failure,
  SUM(osf) as overstrain_failure,
  COUNT(*) as total_count
FROM lgit_pm_bronze
GROUP BY machine_failure