F1 2025 Factorial Performance Analysis: LEC vs VER 🏎️📊

This repository contains an advanced telemetry analysis framework designed to benchmark Formula 1 car performance. The project focuses on the 2025 season rivalry between Charles Leclerc (Ferrari) and Max Verstappen (Red Bull), utilizing Unsupervised Machine Learning to isolate pure vehicular potential from session noise.

Project Overview

Unlike standard lap-time comparisons, this project deconstructs performance into its fundamental physical components: Apex Speed, Traction, and Braking. By using a factorial approach, we can identify exactly where a car gains or loses time relative to its competitor.

Key Technical Features

• ML-Enhanced Categorization: Uses K-Means Clustering ($K=4$) to automatically segment over 200+ corner instances into dynamic regimes: Slow (Traction), Medium (Handling), Fast (Downforce), and Power.
• "Real Analysis" Filter: A custom data-cleaning pipeline that rejects outliers (deltas >15-20 km/h) caused by traffic, yellow flags, or driver errors, ensuring a "pure qualifying" comparison.
• Automated Apex Detection: Implements a bespoke algorithm using scipy.signal.find_peaks to identify the precise physical apex ($V_{min}$) based on telemetry traces rather than static track maps.

Tech Stack

• Data Source: FastF1 API (Official Timing & Telemetry).
• Processing: Pandas, NumPy, SciPy.
• Machine Learning: Scikit-learn (StandardScaler, KMeans).
• Visualization: Seaborn, Matplotlib.