🏎️ Formula 1 World Championship Data Analysis & ML Pipeline

🏁 Project Overview

End-to-end data science project analyzing Formula 1 World Championship data (1950-2020) using machine learning, statistical analysis, and interactive visualizations. Demonstrates complete ML lifecycle from data preprocessing to model optimization achieving 100% accuracy.

🎯 Objective

Predict driver performance and race outcomes using historical F1 data while extracting actionable insights about racing patterns and performance factors.

✨ Key Features & Technical Skills

🔍 Data Engineering

• Multi-source Integration: Merged 6 datasets (25K+ records)
• Feature Engineering: Driver age calculation, performance metrics
• Data Quality: IQR-based outlier detection, missing value imputation
• Statistical Analysis: Skewness testing, correlation analysis

🗺️ Geospatial Analytics

• Interactive Mapping: Folium world map with F1 circuit locations
• GPS Visualization: Custom markers with circuit details

🤖 Machine Learning Pipeline

• 6 Classification Algorithms: Logistic Regression, Random Forest, Decision Tree, KNN, Naive Bayes, SGD
• 3 Scaling Techniques: MinMaxScaler, StandardScaler, RobustScaler
• Performance Optimization: Achieved 40%+ accuracy improvement through scaling

🗂️ Dataset & Performance

📋 Data Sources (1950-2020)

Dataset	Records	Features	Description
results.csv	25,840+	18	Race results and performance metrics
drivers.csv	857	8	Driver biographical information
circuits.csv	77	9	Circuit specifications and coordinates

🎯 Target Variables

• Driver Performance Classification: Based on race finishing positions
• Performance Metrics: Points, lap times, fastest lap speeds

🤖 Machine Learning Models

🔬 Algorithms Implemented

1. Logistic Regression - Linear classification baseline
2. Decision Tree Classifier - Interpretable tree-based model with visualization
3. Random Forest Classifier - Ensemble method for improved accuracy
4. K-Nearest Neighbors (KNN) - Instance-based learning algorithm
5. Gaussian Naive Bayes - Probabilistic classifier
6. Stochastic Gradient Descent (SGD) - Scalable optimization algorithm

⚙️ Feature Scaling Techniques

• MinMaxScaler: Range [0,1] normalization
• StandardScaler: Z-score standardization (μ=0, σ=1)
• RobustScaler: Median and IQR-based scaling (outlier-resistant)

📊 Model Performance

Model	Raw Data	MinMaxScaler	StandardScaler	RobustScaler
Random Forest	95%+	99%+	100%	100%
Decision Tree	90%+	95%+	100%	100%
Logistic Regression	60%	99%+	100%	99%+

📊 Visualization & Analysis

🎨 Advanced Visualizations

• Interactive Correlation Heatmaps: Feature relationship analysis
• Geographic Circuit Mapping: World map with F1 venue locations
• Performance Trend Analysis: Algorithm accuracy comparison charts
• Decision Tree Visualization: Complete model interpretability
• Statistical Distribution Plots: Data quality assessment visualizations

📈 Key Insights Discovered

• Scaling Impact: Demonstrated 40%+ accuracy improvement with proper feature scaling
• Algorithm Performance: Ensemble methods consistently outperformed single algorithms
• Data Quality: Outlier removal significantly improved model generalization
• Feature Importance: Identified critical performance predictors

⚡ Performance Results

Key Achievements:

• ✅ 100% Model Accuracy achieved (Random Forest + StandardScaler)
• ✅ 67% Accuracy Improvement through feature scaling optimization
• ✅ 6 ML Algorithms successfully implemented and compared
• ✅ Interactive Visualizations with geospatial mapping

🛠️ Technologies Used

Core Stack: Python 3.8+ | Pandas | NumPy | Scikit-learn | Jupyter Notebook
Visualization: Matplotlib | Seaborn | Folium (Interactive Maps)
ML Techniques: Classification Algorithms | Feature Scaling | Statistical Analysis

---

⭐ Star this repository if you found it helpful!