preprocessing.json

{
  "context_columns": [
    "race_id",
    "horse_id",
    "track_id",
    "date"
  ],
  "target_column": "is_winner",
  "base_numeric_features": [
    "pace_pressure_score",
    "barrier",
    "early_speed_rating",
    "api_index",
    "weight_delta",
    "days_since_last_run",
    "margin_beaten_last_start",
    "track_condition_win_pct",
    "track_layout_score",
    "distance_mastery_pct",
    "last_600m_rank",
    "sectional_vs_class_avg",
    "jockey_trainer_strike_rate",
    "market_firm_factor"
  ],
  "categorical_columns": [
    "projected_settling_position"
  ],
  "feature_columns": [
    "rel_pace_pressure_score",
    "rel_barrier",
    "rel_early_speed_rating",
    "rel_api_index",
    "rel_weight_delta",
    "rel_days_since_last_run",
    "rel_margin_beaten_last_start",
    "rel_track_condition_win_pct",
    "rel_track_layout_score",
    "rel_distance_mastery_pct",
    "rel_last_600m_rank",
    "rel_sectional_vs_class_avg",
    "rel_jockey_trainer_strike_rate",
    "rel_market_firm_factor",
    "speed_map_score",
    "form_class_score",
    "track_env_score",
    "sectionals_score",
    "human_market_score",
    "expert_prior_score",
    "projected_settling_position"
  ],
  "lower_is_better_features": [
    "barrier",
    "days_since_last_run",
    "last_600m_rank",
    "margin_beaten_last_start"
  ],
  "numeric_fill_medians": {
    "pace_pressure_score": 0.0,
    "barrier": 6.0,
    "early_speed_rating": 55.0,
    "api_index": 4374.53125,
    "weight_delta": 0.0,
    "days_since_last_run": 99.0,
    "margin_beaten_last_start": 1853.0,
    "track_condition_win_pct": 0.1,
    "track_layout_score": 0.65,
    "distance_mastery_pct": 0.0,
    "last_600m_rank": 1854.0,
    "sectional_vs_class_avg": -0.11882353,
    "jockey_trainer_strike_rate": 0.0,
    "market_firm_factor": 0.0
  },
  "macro_buckets": {
    "speed_map": [
      "pace_pressure_score",
      "barrier",
      "early_speed_rating"
    ],
    "form_class": [
      "api_index",
      "weight_delta",
      "days_since_last_run",
      "margin_beaten_last_start"
    ],
    "track_env": [
      "track_condition_win_pct",
      "track_layout_score",
      "distance_mastery_pct"
    ],
    "sectionals": [
      "last_600m_rank",
      "sectional_vs_class_avg"
    ],
    "human_market": [
      "jockey_trainer_strike_rate",
      "market_firm_factor"
    ]
  },
  "macro_weights": {
    "speed_map": 0.25,
    "form_class": 0.25,
    "track_env": 0.2,
    "sectionals": 0.15,
    "human_market": 0.15
  }
}