File size: 12,975 Bytes
46c1c8b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 | """
KRONECTOR β Pre-Race Prediction Data Builder
=============================================
Builds prediction rows for an UPCOMING race using:
- Qualifying sector times + grid positions (from FastF1 qualifying session)
- Practice avg lap times (FP2/FP3)
- Driver form from existing historical parquet
- Championship standings from existing historical parquet
This allows predictions BEFORE the race happens, using only qualifying data.
Usage:
python -m scripts.build_prerace_rows --season 2026 --round 7 --output data_output/prerace_monaco_2026.parquet
python -m scripts.build_prerace_rows --season 2026 --round 7 # saves to data_output/fastf1_races.parquet as temp rows
"""
import argparse
import logging
import os
import sys
from pathlib import Path
import numpy as np
import pandas as pd
import fastf1
import builtins
import requests
import urllib3
builtins.RequestsCookieJar = requests.cookies.RequestsCookieJar
builtins.HTTPAdapter = requests.adapters.HTTPAdapter
builtins.Retry = urllib3.util.Retry
sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
from data.fastf1_pipeline import enable_cache, fetch_qualifying, fetch_practice
logger = logging.getLogger(__name__)
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s | %(name)s | %(levelname)s | %(message)s",
)
def fetch_grid_from_qualifying(season: int, round_num: int) -> pd.DataFrame | None:
"""
Load the qualifying session and extract grid positions + sector times.
Uses lightweight load (laps only, no telemetry/weather/messages) so it works
even when data was just published.
Returns DataFrame with columns: driver_id, driver_name, team, grid_position, circuit_id
"""
enable_cache()
try:
session = fastf1.get_session(season, round_num, "Q")
# Only load laps β skip telemetry, weather, messages (not needed + cause errors on fresh data)
session.load(laps=True, telemetry=False, weather=False, messages=False)
logger.info(f"Qualifying session loaded for {season} R{round_num}")
except Exception as e:
logger.error(f"Could not load qualifying session for {season} R{round_num}: {e}")
logger.error("π‘ If qualifying just finished, wait ~15-30 minutes for FastF1 to process the data, then try again.")
return None
# Try results table first (gives clean classification order)
try:
results = session.results
except Exception:
results = None
if results is not None and not results.empty:
circuit_id = session.event["EventName"]
df = pd.DataFrame({
"driver_id": results["Abbreviation"].values,
"driver_name": results["FullName"].values,
"team": results["TeamName"].values,
"grid_position": range(1, len(results) + 1),
"circuit_id": circuit_id,
"quali_status": results["Status"].values if "Status" in results.columns else "Finished",
})
logger.info(f"β
Grid from qualifying results: {season} R{round_num} ({circuit_id}) β {len(df)} drivers")
logger.info(f" P1: {df.iloc[0]['driver_name']} ({df.iloc[0]['team']})")
return df
# Fallback: derive grid from laps (best lap time ordering)
try:
laps = session.laps
except Exception:
laps = None
if laps is None or laps.empty:
logger.warning(f"No qualifying laps available for {season} R{round_num}")
logger.warning("π‘ Data may not be available yet β try again in 15-30 minutes.")
return None
circuit_id = session.event["EventName"]
# Best lap per driver β sort ascending to get grid order
best = laps.groupby("Driver")["LapTime"].min().sort_values().reset_index()
# Map driver abbreviation to full name/team via session results if available
driver_info = {}
if session.results is not None and not session.results.empty:
for _, r in session.results.iterrows():
driver_info[r["Abbreviation"]] = {
"driver_name": r.get("FullName", r["Abbreviation"]),
"team": r.get("TeamName", "Unknown"),
}
records = []
for pos, (_, row) in enumerate(best.iterrows(), start=1):
abbr = row["Driver"]
info = driver_info.get(abbr, {"driver_name": abbr, "team": "Unknown"})
records.append({
"driver_id": abbr,
"driver_name": info["driver_name"],
"team": info["team"],
"grid_position": pos,
"circuit_id": circuit_id,
"quali_status": "Finished",
})
df = pd.DataFrame(records)
logger.info(f"β
Grid from lap times (fallback): {season} R{round_num} ({circuit_id}) β {len(df)} drivers")
logger.info(f" P1: {df.iloc[0]['driver_name']} ({df.iloc[0]['team']})")
return df
def get_driver_context(existing_df: pd.DataFrame, season: int, round_num: int) -> pd.DataFrame:
"""
Pull driver_form_last3 and championship_standing for each driver
from the most recent data available before this round.
"""
# Get data up to but NOT including this round/season
prior = existing_df[
(existing_df["season"] < season) |
((existing_df["season"] == season) & (existing_df["round"] < round_num))
].copy()
if prior.empty:
logger.warning("No prior data found for driver context β using defaults")
return pd.DataFrame(columns=["driver_id", "driver_form_last3", "championship_standing"])
# Latest championship standing per driver this season (or prior season)
context_rows = []
for driver_id in prior["driver_id"].unique():
driver_data = prior[prior["driver_id"] == driver_id].sort_values(
["season", "round"], ascending=True
)
if driver_data.empty:
continue
# driver_form_last3: mean finish position of last 3 races
recent = driver_data.tail(3)
form = recent["finish_position"].mean() if "finish_position" in recent.columns else np.nan
# championship_standing: most recent value
standing = np.nan
if "championship_standing" in driver_data.columns:
valid = driver_data["championship_standing"].dropna()
if not valid.empty:
standing = valid.iloc[-1]
# safety_car_probability: get from the circuit if available
context_rows.append({
"driver_id": driver_id,
"driver_form_last3": form,
"championship_standing": standing,
})
return pd.DataFrame(context_rows)
def build_prerace_rows(
season: int,
round_num: int,
existing_parquet: str = "data_output/fastf1_races.parquet",
) -> pd.DataFrame | None:
"""
Build a DataFrame of pre-race prediction rows for an upcoming race.
These rows use qualifying + practice data for features, and set:
- finish_position = NaN (unknown β will be imputed/excluded)
- win_probability target = 0 (dummy β not used in inference)
- telemetry_available = True
Returns a DataFrame compatible with ml.predict.predict_dataframe()
"""
# 1. Get grid from qualifying results
grid_df = fetch_grid_from_qualifying(season, round_num)
if grid_df is None:
return None
# 2. Get qualifying sector times
quali_sectors = fetch_qualifying(season, round_num)
if quali_sectors is not None:
grid_df = grid_df.merge(quali_sectors, on="driver_id", how="left")
else:
logger.warning("No qualifying sector times β using NaN")
grid_df["sector_1_time"] = np.nan
grid_df["sector_2_time"] = np.nan
grid_df["sector_3_time"] = np.nan
# 3. Get practice lap times
practice_df = fetch_practice(season, round_num)
if practice_df is not None:
grid_df = grid_df.merge(practice_df, on="driver_id", how="left")
else:
grid_df["avg_lap_time_practice"] = np.nan
# 4. Pull driver form + championship standing from existing historical data
if Path(existing_parquet).exists():
existing_df = pd.read_parquet(existing_parquet)
context_df = get_driver_context(existing_df, season, round_num)
grid_df = grid_df.merge(context_df, on="driver_id", how="left")
else:
logger.warning(f"No existing parquet at {existing_parquet} β driver context will be NaN")
grid_df["driver_form_last3"] = np.nan
grid_df["championship_standing"] = np.nan
# 5. Fill in race-level fields that won't be known until after the race
grid_df["season"] = season
grid_df["round"] = round_num
grid_df["finish_position"] = np.nan # Unknown β will be imputed
grid_df["tire_compound"] = np.nan
grid_df["tire_age_laps"] = np.nan
grid_df["fresh_tire"] = np.nan
grid_df["pit_stop_count"] = np.nan
grid_df["team_pit_speed"] = np.nan
grid_df["weather_temp_track"] = np.nan
grid_df["weather_rainfall"] = np.nan
grid_df["telemetry_available"] = True
# safety_car_probability: use historical average for this circuit if available
if Path(existing_parquet).exists():
existing_df = pd.read_parquet(existing_parquet)
circuit_name = grid_df["circuit_id"].iloc[0]
circuit_data = existing_df[
existing_df["circuit_id"].astype(str).str.contains(
circuit_name.split()[0], case=False, na=False
)
]
if not circuit_data.empty and "safety_car_probability" in circuit_data.columns:
sc_prob = circuit_data["safety_car_probability"].dropna().mean()
grid_df["safety_car_probability"] = sc_prob if not np.isnan(sc_prob) else 0.0
else:
grid_df["safety_car_probability"] = 0.0
# pole_conversion_rate: historical pole-to-win % at this circuit
if not circuit_data.empty and "finish_position" in circuit_data.columns:
poles = circuit_data[circuit_data["grid_position"] == 1].dropna(subset=["finish_position"])
if not poles.empty:
pcr = (poles["finish_position"] == 1).mean()
grid_df["pole_conversion_rate"] = pcr
logger.info(f" Pole conversion rate at {circuit_name}: {pcr:.1%}")
else:
grid_df["pole_conversion_rate"] = 0.5
else:
grid_df["pole_conversion_rate"] = 0.5
# career_race_starts: count driver's past races
for idx, row in grid_df.iterrows():
d_id = row["driver_id"]
starts = len(existing_df[existing_df["driver_id"] == d_id])
grid_df.at[idx, "career_race_starts"] = starts
else:
grid_df["safety_car_probability"] = 0.0
grid_df["pole_conversion_rate"] = 0.5
grid_df["career_race_starts"] = 0
logger.info(f"\nβ
Pre-race rows built for {season} R{round_num} β {len(grid_df)} drivers")
logger.info(f" Grid P1: {grid_df.iloc[0]['driver_name']} (qualifying pole)")
logger.info("\nGrid order:")
for _, row in grid_df.iterrows():
s1 = f"{row.get('sector_1_time', np.nan):.3f}s" if pd.notna(row.get('sector_1_time')) else "N/A"
logger.info(f" P{int(row['grid_position'])}: {row['driver_name']} ({row['team']}) | S1={s1}")
return grid_df
def main():
parser = argparse.ArgumentParser(description="KRONECTOR β Pre-Race Row Builder")
parser.add_argument("--season", type=int, required=True)
parser.add_argument("--round", type=int, required=True, dest="round_num")
parser.add_argument(
"--output",
type=str,
default=None,
help="Output parquet path. Defaults to data_output/prerace_{circuit}_{season}.parquet",
)
parser.add_argument(
"--existing",
type=str,
default="data_output/fastf1_races.parquet",
help="Existing historical parquet for driver form/standings context",
)
args = parser.parse_args()
df = build_prerace_rows(
season=args.season,
round_num=args.round_num,
existing_parquet=args.existing,
)
if df is None:
logger.error("Failed to build pre-race rows. Exiting.")
sys.exit(1)
# Save output β default to data_output/prerace/ so the API auto-loads it on startup
circuit_slug = df["circuit_id"].iloc[0].replace(" ", "_").lower()[:20] if not df.empty else "unknown"
out_path = args.output or f"data_output/prerace/prerace_{circuit_slug}_{args.season}.parquet"
Path(out_path).parent.mkdir(parents=True, exist_ok=True)
df.to_parquet(out_path, index=False)
logger.info(f"\nπΎ Saved to: {out_path}")
print(f"\nPre-race parquet saved: {out_path}")
print(f"Rows: {len(df)} drivers")
print(f"\nTo run predictions:")
print(f" Use the API: POST /predict/f1 with data_path pointing to this file")
print(f" Or: python -m ml.predict --run-id <YOUR_RUN_ID> --data-path {out_path}")
if __name__ == "__main__":
main()
|