Titanic - Machine Learning from Disaster 🛳️

This repository contains three progressively improved models submitted to Kaggle's Titanic survival prediction competition:

🔧 Models Included

1. Baseline Random Forest – default sklearn settings
2. Tuned Random Forest – GridSearchCV optimized
3. Tuned XGBoost – grid search + feature engineering

📁 Structure

• notebooks/ – All model development notebooks
• submissions/ – CSVs ready for Kaggle submission
• requirements.txt – Python packages used

📊 Results

Model	CV Accuracy	Kaggle Rank
Baseline RF	~82.1%	#14,329
Tuned RF	~83.7%	#10,729
Tuned XGBoost	~84.9%	#9,686

🚀 Features Used

• Title & Deck extracted from Name/Cabin
• FamilySize, IsAlone
• Pclass, Age, Sex, Fare, Embarked

📌 Future Work

• Ensemble Voting Classifier
• LightGBM trials
• Additional feature engineering

---