{
    "table_name": "bedtime_screentime_sleep_debt",
    "table_type_tags": [
        {
            "tag": "dimension",
            "confidence": 0.9
        }
    ],
    "primary_type": "dimension",
    "insufficient_evidence": false,
    "entity": {
        "columns": [
            "user_id"
        ],
        "unique_per_row": true
    },
    "time_axis": null,
    "reasoning": "user_id has Uniqueness 1.000 (8,500 unique values, RPUV 1.00) and names the thing each row is about, so the entity is unique per row. There is no event timestamp or sequence column. The only datetime column is _ingestion_timestamp, which is provenance and ignored. There is no lifecycle in columns and no repeating parent container, so the row is one profile per user with attributes (age, chronotype, occupation_type) and one night's summary measures (total_sleep_hours, next_day_fatigue_score, sleep_debt_category). That makes it a dimension."
}