F1 Race Insights

Machine Learning for Formula 1 Predictions

WHAT IS THIS PROJECT?

An AI System That Predicts Race Outcomes

This is a machine learning application that predicts winners of Formula 1 car races. Think of it like a sports betting algorithm, but for racing. The system analyzes historical data from thousands of past races to predict future outcomes with 98.7% accuracy.

No F1 knowledge required! This page explains everything you need to understand what this project demonstrates technically.

Technical Achievements

98.7%
AUC-ROC Score
Our models correctly rank winners 98.7% of the time (random chance would be 50%)
8
ML Models
We built 8 different prediction systems and compared their performance
10,000
Simulations
We run 10,000 possible season outcomes to estimate championship probabilities
68
Features
Our AI considers 68 different factors when making each prediction

Key Features to Explore

Each feature is explained in plain English - no F1 knowledge needed

Season Simulator

Simulates entire racing seasons using machine learning

💡 Plain English:

Like predicting who wins the Super Bowl, but for 24 car races in a season. We run 10,000 simulations to see all possible outcomes.

Monte Carlo simulation, probabilistic modeling, what-if analysis
Explore feature

Model Architecture

Shows all 8 machine learning models used for predictions

💡 Plain English:

We built 8 different AI systems that each predict race winners in their own way, then compare their accuracy.

XGBoost, Neural Networks, Ensemble methods, model comparison
Explore feature

SHAP Explainer

Explains WHY the AI made each prediction

💡 Plain English:

Most AI is a "black box" - you don't know why it decided something. Our system shows exactly which factors influenced each prediction.

SHAP values, feature importance, model interpretability
Explore feature

What-If Lab

Changes historical events to see alternate outcomes

💡 Plain English:

What if a controversial decision in a famous race went differently? This tool rewrites history to show what could have happened.

Causal inference, counterfactual analysis, domain simulation
Explore feature

Live Predictions

Real-time race predictions as events unfold

💡 Plain English:

Like a sports prediction app that updates every second during a game, adjusting odds based on what's happening.

Real-time inference, streaming data, live probability updates
Explore feature

Backtest Results

Tests how accurate our predictions are on past data

💡 Plain English:

We test our AI against races that already happened (without cheating by using future data) to prove it actually works.

Walk-forward validation, AUC-ROC, cross-validation
Explore feature

Quick F1 Glossary (Optional)

If you want to understand some F1 terminology you might see in the app:

Formula 1 (F1)
The world's most prestigious car racing series - 20 drivers, 10 teams, 24 races per year
Grand Prix
A single race in the F1 calendar, held in different countries (like playoffs in other sports)
Championship
Season-long competition where drivers earn points in each race; most points wins
Pit Stop
When a car briefly stops during a race to change tires or make repairs
Machine Learning
Teaching computers to find patterns in data and make predictions automatically

Technical Skills Demonstrated

Machine Learning

  • • 8 different ML models (XGBoost, CatBoost, Neural Networks)
  • • Custom neural architecture (NBT-TLF)
  • • Feature engineering with 68 predictors
  • • SHAP explainability
  • • Monte Carlo simulations

Data Engineering

  • • ETL pipeline for F1 race data
  • • Real-time data streaming
  • • Walk-forward backtesting
  • • Feature store implementation
  • • API rate limiting and caching

Full-Stack Development

  • • Next.js 14 with React
  • • TypeScript throughout
  • • Python ML backend (FastAPI)
  • • Responsive, accessible UI
  • • Interactive data visualizations

MLOps & DevOps

  • • Model versioning and registry
  • • A/B testing infrastructure
  • • Model drift monitoring
  • • Docker containerization
  • • CI/CD pipeline

Ready to Explore?

Start with the Season Simulator to see ML predictions in action

Start SimulatingView ML ModelsBrowse All Features