The lap_times.csv table contains individual lap times for every driver in each race, providing detailed telemetry for race analysis.
Schema
| Field | Type | Description |
|---|
raceId | integer | Foreign key to races.csv |
driverId | integer | Foreign key to drivers.csv |
lap | integer | Lap number |
position | integer | Driver’s position at the end of this lap |
time | string | Lap time in MM:SS.mmm format |
milliseconds | integer | Lap time in milliseconds for precise calculations |
This table uses a composite key of (raceId, driverId, lap) rather than a single unique identifier.
Sample Data
| raceId | driverId | lap | position | time | milliseconds |
|---|
| 841 | 20 | 1 | 1 | 1:38.109 | 98109 |
| 841 | 20 | 2 | 1 | 1:33.006 | 93006 |
| 841 | 20 | 3 | 1 | 1:32.713 | 92713 |
| 841 | 20 | 4 | 1 | 1:32.803 | 92803 |
Relationships
References:
lap_times.raceId → races.raceIdlap_times.driverId → drivers.driverId
Dataset Statistics
- Total Records: 616,741 lap times
- Date Range: Lap-by-lap data available from 1996 onwards
- Laps per Race: Varies by circuit (typically 50-70 laps)
Example Queries
Get all laps for a specific race
import pandas as pd
lap_times = pd.read_csv('lap_times.csv')
race_laps = lap_times[lap_times['raceId'] == 841]
print(race_laps.head(20))
Find fastest lap in a race
race_laps = lap_times[lap_times['raceId'] == 1000]
fastest = race_laps.nsmallest(1, 'milliseconds')
print(fastest[['driverId', 'lap', 'time', 'milliseconds']])
Analyze driver pace throughout a race
import matplotlib.pyplot as plt
driver_laps = lap_times[
(lap_times['raceId'] == 1000) &
(lap_times['driverId'] == 1)
]
plt.plot(driver_laps['lap'], driver_laps['milliseconds'] / 1000)
plt.xlabel('Lap Number')
plt.ylabel('Lap Time (seconds)')
plt.title('Driver Pace Analysis')
plt.show()
Calculate average lap time by driver
race_laps = lap_times[lap_times['raceId'] == 1000]
average_times = race_laps.groupby('driverId')['milliseconds'].mean()
print(average_times.sort_values())
Identify pit stop laps
# Pit stop laps are typically 20-30+ seconds slower
driver_laps = lap_times[
(lap_times['raceId'] == 1000) &
(lap_times['driverId'] == 1)
].copy()
driver_laps['diff'] = driver_laps['milliseconds'].diff()
pit_laps = driver_laps[driver_laps['milliseconds'] > driver_laps['milliseconds'].mean() * 1.3]
print(pit_laps[['lap', 'time']])
Compare lap times between drivers
driver1_laps = lap_times[
(lap_times['raceId'] == 1000) &
(lap_times['driverId'] == 1)
].set_index('lap')
driver2_laps = lap_times[
(lap_times['raceId'] == 1000) &
(lap_times['driverId'] == 20)
].set_index('lap')
comparison = pd.DataFrame({
'driver1': driver1_laps['milliseconds'],
'driver2': driver2_laps['milliseconds']
})
comparison['gap'] = comparison['driver2'] - comparison['driver1']
print(comparison)
Detect safety car periods
# Safety car laps show all drivers with similar slow times
race_laps = lap_times[lap_times['raceId'] == 1000]
avg_by_lap = race_laps.groupby('lap')['milliseconds'].mean()
std_by_lap = race_laps.groupby('lap')['milliseconds'].std()
# Low variance + slow times = potential safety car
safety_car_laps = avg_by_lap[
(std_by_lap < std_by_lap.mean() * 0.5) &
(avg_by_lap > avg_by_lap.mean() * 1.1)
]
print(f"Potential safety car laps: {list(safety_car_laps.index)}")
Calculate race pace degradation
# Compare early stint vs late stint pace
driver_laps = lap_times[
(lap_times['raceId'] == 1000) &
(lap_times['driverId'] == 1)
]
early_laps = driver_laps[driver_laps['lap'].between(2, 10)]
late_laps = driver_laps[driver_laps['lap'].between(40, 50)]
early_avg = early_laps['milliseconds'].mean()
late_avg = late_laps['milliseconds'].mean()
degradation = ((late_avg - early_avg) / early_avg) * 100
print(f"Pace degradation: {degradation:.2f}%")
Notes
- Lap time data is most complete from 1996 onwards
- First lap times are typically slower due to standing start
- Pit stop laps show significantly slower times (include pit lane time)
- Safety car and VSC periods affect lap times considerably
- Track evolution means later laps are often faster (unless tire degradation)
milliseconds field is preferred for calculations to avoid parsing time strings- Outlier lap times may indicate incidents, pit stops, or data errors
- Position changes between laps help track overtakes
