{
    "action": "final",
    "content": "## Dataset Overview\n\nThis dataset is about **professional boxing matches and fighter statistics**. It contains two distinct tables:\n\n### What a Single Row Represents\n\n**In `raw.fighters` (2,760 rows):**\nEach row represents a **single professional boxer** with their career statistics and physical attributes. For example:\n- **Saul Alvarez** (Canelo): 54 wins, 1 loss, 2 draws, 63.2% KO rate, Orthodox stance, age 32, from Mexico\n- **Azizbek Abdumuxtar Abdugofurov**: 0-0-0 record, Orthodox stance, from Uzbekistan\n\nThe fighters table captures:\n- Career record (wins, losses, draws)\n- Fighting style (stance)\n- Physical measurements (height, reach)\n- Demographics (age, country)\n- Knockout effectiveness (ko_rate)\n\n**In `raw.popular_matches` (152 rows):**\nEach row represents a **single high-profile boxing match** between two opponents, with detailed pre-fight metrics and the outcome. For example:\n- **Vasyl Lomachenko vs Luke Campbell** (31 August 2019): Lomachenko won via UD in round 12\n- **Gennady Golovkin vs Sergiy Derevyanchenko** (5 October 2019): Golovkin won via UD in round 12\n\nThe matches table captures:\n- Match details (date, place, opponents)\n- Comparative fighter metrics (estimated punch power, punch resistance, ability to take punch)\n- Historical performance (rounds boxed, KO percentages, average weight)\n- Match outcome (verdict)\n\n### Dataset Purpose\n\nBased on the dataset name (\"predict-winner\") and the rich comparative metrics in the matches table, this dataset appears designed for **predictive modeling of boxing match outcomes**. The matches table provides paired opponent statistics that could be used to train models predicting which fighter will win based on their physical attributes, fighting history, and estimated abilities."
}