```json { "classifications": [ { "column": "dataset_id", "role": "identifier", "confidence": 1.0, "reason": "Primary key with 'id' suffix, used to identify datasets; 99 distinct values for 198 rows indicates it's a foreign key joining to a dataset table" }, { "column": "batch_id", "role": "identifier", "confidence": 1.0, "reason": "Primary/unique key with 'id' suffix and 198 distinct values for 198 rows; identifies each batch uniquely" }, { "column": "tenant_id", "role": "identifier", "confidence": 1.0, "reason": "Foreign key with 'id' suffix referencing tenant; used for joins even though only 1 distinct value in this dataset" }, { "column": "submitted_at", "role": "dimension", "confidence": 0.95, "reason": "Timestamp used for filtering, grouping by time periods, and temporal analysis" }, { "column": "insights_ready_at", "role": "dimension", "confidence": 0.95, "reason": "Timestamp used for filtering, grouping by time periods, and temporal analysis" }, { "column": "time_to_insights_ms", "role": "measure", "confidence": 1.0, "reason": "Numeric duration in milliseconds that would be aggregated (avg, sum, min, max) to analyze performance" }, { "column": "status", "role": "dimension", "confidence": 1.0, "reason": "Categorical field with 6 distinct values (backfilled, complete, degraded, failed, in_progress); used for grouping and filtering" }, { "column": "status_reason", "role": "attribute", "confidence": 1.0, "reason": "Free-form descriptive text explaining status; displayed but not grouped or aggregated" }, { "column": "phases_json", "role": "attribute", "confidence": 0.9, "reason": "JSON field for descriptive content; 100% null but intended for display, not aggregation" }, { "column": "created_at", "role": "dimension", "confidence": 0.95, "reason": "Timestamp used for filtering, grouping by time periods, and temporal analysis" }, { "column": "updated_at", "role": "dimension", "confidence": 0.95, "reason": "Timestamp used for filtering, grouping by time periods, and temporal analysis" }, { "column": "insights_started_at", "role": "dimension", "confidence": 0.95, "reason": "Timestamp used for filtering, grouping by time periods, and temporal analysis" } ] } ```