Skip to main content
Race pace analysis is one of the most common tasks in F1 data science. It involves looking at lap times over a long stint to see which driver is faster and how well they manage their tires.

​
Loading the Data

For this analysis, we’ll look at the 2024 Abu Dhabi Grand Prix. We’ll use the polars lib for faster filtering. 
import tif1
import matplotlib.pyplot as plt
import seaborn as sns

# Load the session with polars lib
session = tif1.get_session(2024, "Abu Dhabi Grand Prix", "Race", lib="polars")
laps = session.laps

​
Cleaning the Data

In a race, lap times can be skewed by:
  • Safety Car periods (very slow laps)
  • Pit stops (very slow laps)
  • Lap 1 (standing start)
We need to filter these out to get a “true” race pace. 
# Convert to pandas for easier plotting
df = laps.to_pandas()

# Filter out very slow laps and pit laps
# Typically, race laps are within a certain range
clean_laps = df[
    (df["LapTime"] < df["LapTime"].min() * 1.07) & # Within 7% of fastest lap
    (df["PitInTime"].isna()) &
    (df["PitOutTime"].isna()) &
    (df["LapNumber"] > 1)
]

​
Comparing Drivers

Let’s compare the pace of the top two finishers. 
# Filter for specific drivers
drivers = ["VER", "LEC"]
comparison_df = clean_laps[clean_laps["Driver"].isin(drivers)]

# Create a boxplot to see distribution of lap times
plt.figure(figsize=(10, 6))
sns.boxplot(x="Driver", y="LapTime", data=comparison_df, palette="viridis")
plt.title("Race Pace Comparison: VER vs LEC")
plt.ylabel("Lap Time (s)")
plt.show()

​
Interpreting the Results

  • Median (line in the box): Represents the typical race pace.
  • Box size (IQR): Represents consistency. A smaller box means the driver was more consistent.
  • Whiskers/Outliers: Show variance caused by traffic or minor errors.

​
Visualizing pace over the race

To see how the pace evolved (tire degradation), we can plot lap times against lap numbers. 
plt.figure(figsize=(12, 6))
for driver in drivers:
    driver_laps = comparison_df[comparison_df["Driver"] == driver]
    plt.plot(driver_laps["LapNumber"], driver_laps["LapTime"], label=driver)

plt.title("Lap Times Evolution")
plt.xlabel("Lap Number")
plt.ylabel("Lap Time (s)")
plt.legend()
plt.grid(True, alpha=0.3)
plt.show()

​
Conclusion

By following this workflow, you can:
  1. Load high-quality timing data.
  2. Clean it to remove anomalies.
  3. Compare drivers using statistical distributions.
  4. Visualize performance trends across the entire race.