```json { "table_name": "samples", "primary_key": "id", "candidates": [ { "rank": 0, "partition_key": "tables_catalog_id", "partition_key_label": "primary_identifier", "partition_key_reasoning": "RPUV of 19.92 indicates ~20 rows per catalog table, which is ideal for partitioning. This column groups sample rows by their source table, making it the natural organizer for this metadata table. Top 3 coverage of 0.59% confirms good distribution across 508 unique tables.", "clustering_key": null, "clustering_key_label": null, "clustering_key_reasoning": "Dimension tables typically do not require clustering. The id column already provides unique ordering, and there is no temporal or sequential axis to sort within each partition.", "world_line": null, "session_column": null, "confidence": "HIGH", "notes": "This is a dimension table storing sample data rows from various catalog tables. The id column is the unique identifier (RPUV=1.00, uniqueness=1.000). Partitioning by tables_catalog_id groups samples by their source table, which is the primary organizational axis. The row_json column contains the actual sample data payload. The _ingestion_timestamp and _batch_id are metadata columns with no variance (all rows share the same value), so they cannot serve as partition or clustering keys." } ] } ```