{
    "table_name": "return_kaggle",
    "table_type_tags": [
        {
            "tag": "event",
            "confidence": 0.78
        },
        {
            "tag": "dimension",
            "confidence": 0.2
        }
    ],
    "primary_type": "event",
    "insufficient_evidence": false,
    "entity": {
        "columns": [
            "Name",
            "Date",
            "Tournament",
            "Rd"
        ],
        "unique_per_row": false
    },
    "time_axis": {
        "columns": [
            "Date",
            "Rd"
        ],
        "kind": "event_time"
    },
    "reasoning": "Each row is one match played by a player. The match is identified by Name (462 unique, RPUV 513.41) plus Date (1,424 unique), Tournament and Rd, and the `against` opponent column is 100% empty. A row exists because a match happened. Date plus the round sequence (R128 → F) orders a player's matches, but the per-match return stats (TPW, RPW, BPCnv, Time) do not accumulate or replay into a player state, so this is bare occurrences rather than path. Rows are not clock-driven samples, so not state. A weak dimension reading remains because a match-stats record is close to a catalog of occurrences, though the repeating Name actor argues against it."
}