Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 

Repository files navigation

Motorsport Performance Analytics & Strategy Insights (Formula 1) — Exploratory Data Analysis (1950–2024)

A comprehensive exploratory data analysis of 75 years of Formula 1 racing — covering 1,100+ races, 800+ drivers, and 200+ constructors across 9 interconnected datasets. The goal was to move beyond surface-level leaderboards and uncover the structural patterns that actually determine performance, consistency, and dominance in F1.

Key Findings

Driver Performance

Metric Finding
All-Time Points Leader Lewis Hamilton — 4,820.5 pts across career
Most Efficient Winner Juan Fangio — 41.4% win rate among regular competitors
Most Consistent Front Runner Lewis Hamilton — 5.55 std dev in finishing position
Closest Championship Ever 1984: Niki Lauda beat Alain Prost by just 0.5 points
Best Single Season Max Verstappen 2023 — 530 points (highest in modern era)

Insight: Hamilton's dominance is not just volumetric — he leads in both raw output and consistency metrics, suggesting sustained excellence rather than era-specific advantage. Fangio's win ratio remains unmatched despite competing in an era with 7–9 races per season and DNF rates exceeding 60%.

Constructor Intelligence

Metric Finding
Most Wins (All-Time) Ferrari — 249 wins across 7 decades
Hybrid Era Dominance Mercedes — 129 wins, Red Bull — 122 wins
Most Reliable Brawn GP — lowest DNF ratio in dataset
Biggest YoY Improvement Mercedes 2014 — +341 points (hybrid regulation shift)
Street Circuit Benchmark Brawn GP — 2.75 avg finish on street circuits

Insight: Ferrari's lead reflects longevity, not peak dominance — Mercedes and Red Bull compressed comparable win counts into fewer seasons. Brawn's 2009 season stands as the most efficient single-season performance in the dataset: highest win ratio, lowest DNF rate, near-perfect qualifying-to-finish conversion.

Circuit & Strategic Intelligence

Qualifying matters — but less than expected

  • Pole-to-win conversion rate: 42.8% — meaningful but far from deterministic
  • Qualifying-to-finish correlation: r = 0.446 — grid position explains ~20% of race outcome variance
  • Circuits amplify this effect: Yas Marina (68.8%), Marina Bay (66.7%), Barcelona (65.4%) heavily reward front-row starts due to limited overtaking geometry

Pit stop strategy is widely misunderstood

  • Pit stop count vs finish position: r = 0.039 (near-zero correlation)
  • Conclusion: when and why you pit matters far more than how many times or how fast — tire strategy and track position dominate pit stop speed

Circuit character shapes outcomes

Circuit Trait Example Implication
Fastest avg lap Red Bull Ring (74.2s) Low-downforce setup critical
Highest variance Mugello (60,324 ms²) Weather/safety car sensitive
Hardest to overtake Istanbul / Bahrain / Shanghai Qualifying premium elevated
Most unpredictable Sochi (37.5% unique winner ratio) Strategy lottery

Historical Reliability Trend

F1 reliability has undergone a 5–6x structural improvement over 75 years:

Era Avg DNF Rate Character
1980s ~60–70% Mechanical chaos, attrition racing
1990s ~45–50% Improving but volatile
2000s ~30% Modern standards emerging
2010s ~18% Hybrid reliability era
2020s ~12–15% Engineering maturity

This is not just an engineering story — it fundamentally changed race strategy. When DNF rates were high, attrition was the strategy. Modern F1 is won on pace and pit timing, not survival.

Project Structure


├── Formula1_EDA.ipynb                          # Main analysis notebook
├── README.md                         # Project overview
│
└── formula-1-world-championship-1950-2020.zip                        # Dataset (9 CSV files)

Analysis Sections

Section Key Questions Answered
Driver Analysis Leaderboards, win ratios, consistency scoring, circuit specialization, street circuit performance
Constructor Analysis Team dominance, YoY improvement, qualifying vs race conversion, DNF reliability
Circuit Analysis Race frequency, lap time distribution, overtaking difficulty, pole position advantage, predictability scoring
Qualifying & Race Performance Pole-to-win rate, grid-finish correlation, position conversion efficiency by segment
Pit Stop Analysis Speed benchmarks, consistency scoring, strategic impact on race outcome
Reliability & Attrition DNF trends by driver/constructor/circuit/era, historical reliability evolution
Season & Historical Trends Calendar growth, lap time evolution, championship competitiveness, dominance ratios

Methodology Notes

Outlier handling: Driver comparisons use IQR-based filtering to exclude extremely short or long careers from ratio metrics (win rate, avg finish), ensuring comparisons reflect sustained competitive participation rather than single-race anomalies.

Constructor filtering is done on the basis of participation in atleast one season.

Segmentation: Drivers and constructors are segmented into Front Runners (avg qualifying position ≤ 5) and Midfield (avg qualifying position 5–15) before consistency analysis — because a 3.0 std dev means something very different for a race winner vs a points scorer.

DNF classification: Any result status not containing "Finished" or "+N laps" is classified as a DNF, capturing mechanical failures, accidents, and disqualifications.

Dataset

Source: Formula 1 World Championship (1950–2020) — Kaggle

Note : Even though the dataset is named as "1950-2020", it contains data till 2024.

Coverage: 1950–2024 | 9 relational CSV files | 1,100+ races

Libraries Used

python
pandas
numpy
matplotlib
seaborn

About

Exploratory analysis of 75 years of F1 racing — drivers, constructors, circuits, strategy & reliability across 1,100+ races (1950–2024)

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages