models/metadata.json

{
  "best_model_name": "RF (n=300)",
  "feature_columns": [
    "grid_position",
    "driver_id_enc",
    "constructor_id_enc",
    "circuit_id_enc",
    "season",
    "round",
    "temp_max",
    "precipitation",
    "windspeed_max",
    "driver_avg_change",
    "driver_circuit_avg",
    "constructor_avg_change",
    "is_pitlane_start",
    "driver_dnf_rate",
    "constructor_dnf_rate",
    "championship_position",
    "quali_time_sec",
    "quali_gap_to_pole",
    "quali_speed_fl",
    "driver_race_pace"
  ],
  "label_map": {
    "Gained": 0,
    "Held": 1,
    "Lost": 2
  },
  "train_seasons": [
    2020,
    2021,
    2022,
    2023
  ],
  "val_seasons": [
    2024
  ],
  "test_seasons": [
    2025
  ],
  "split_strategy": "season-based (no random split)",
  "imputer": "SimpleImputer(strategy=median) inside Pipeline",
  "class_weight": "balanced (RF only)",
  "threshold_config": {
    "Lost": 1.0
  },
  "iteration_summary": [
    {"iteration": "1 – Baseline",    "model": "Logistic Regression",   "cv_f1": 0.3605, "cv_std": 0.0225, "cv_acc": 0.3942},
    {"iteration": "1 – Baseline",    "model": "RF (n=100)",             "cv_f1": 0.5380, "cv_std": 0.0198, "cv_acc": 0.5600},
    {"iteration": "2 – Weather",     "model": "Logistic Regression",   "cv_f1": 0.3624, "cv_std": 0.0171, "cv_acc": 0.3888},
    {"iteration": "2 – Weather",     "model": "RF (n=200)",             "cv_f1": 0.5275, "cv_std": 0.0180, "cv_acc": 0.5542},
    {"iteration": "2 – Weather",     "model": "XGBoost (n=200)",        "cv_f1": 0.5358, "cv_std": 0.0175, "cv_acc": 0.5439},
    {"iteration": "3 – History",     "model": "RF (n=300)",             "cv_f1": 0.5358, "cv_std": 0.0184, "cv_acc": 0.5772},
    {"iteration": "3 – History",     "model": "XGBoost (n=300)",        "cv_f1": 0.5406, "cv_std": 0.0304, "cv_acc": 0.5523},
    {"iteration": "3 – History",     "model": "GradBoost (n=200)",      "cv_f1": 0.5551, "cv_std": 0.0241, "cv_acc": 0.5772},
    {"iteration": "4 – Quali/Pace",  "model": "RF (n=300)",             "cv_f1": 0.5200, "cv_std": 0.0183, "cv_acc": 0.5642},
    {"iteration": "4 – Quali/Pace",  "model": "XGBoost (n=300)",        "cv_f1": 0.5414, "cv_std": 0.0310, "cv_acc": 0.5535},
    {"iteration": "4 – Quali/Pace",  "model": "GradBoost (n=200)",      "cv_f1": 0.5497, "cv_std": 0.0204, "cv_acc": 0.5718}
  ],
  "results": {
    "RF (n=300)": {
      "val_2024": {
        "accuracy": 0.5658,
        "f1_weighted": 0.5517
      },
      "test_2025": {
        "accuracy": 0.5658,
        "f1_weighted": 0.5396
      }
    },
    "XGBoost (n=300)": {
      "val_2024": {
        "accuracy": 0.428,
        "f1_weighted": 0.4468
      },
      "test_2025": {
        "accuracy": 0.5094,
        "f1_weighted": 0.515
      }
    },
    "GradBoost (n=200)": {
      "val_2024": {
        "accuracy": 0.4885,
        "f1_weighted": 0.5039
      },
      "test_2025": {
        "accuracy": 0.5115,
        "f1_weighted": 0.5177
      }
    }
  },
  "best_model_final": {
    "val_2024": {
      "baseline": {
        "accuracy": 0.5658,
        "f1_weighted": 0.5517,
        "f1_macro": 0.4656
      },
      "threshold": {
        "accuracy": 0.5658,
        "f1_weighted": 0.5517,
        "f1_macro": 0.4656
      }
    },
    "test_2025": {
      "baseline": {
        "accuracy": 0.5658,
        "f1_weighted": 0.5396,
        "f1_macro": 0.4872
      },
      "threshold": {
        "accuracy": 0.5658,
        "f1_weighted": 0.5396,
        "f1_macro": 0.4872
      }
    }
  }
}