Titanic Survival Prediction

A machine learning project to predict passenger survival on the Titanic using the Titanic dataset. This repository preprocesses the data, engineers new features, trains multiple models, and optimizes the best-performing model to achieve high accuracy.

---

Project Overview

This project demonstrates a complete machine learning pipeline:

• Data Preprocessing: Handling missing values, encoding categorical variables, and normalizing numerical features.
• Feature Engineering: Creating meaningful features like FamilySize and IsAlone.
• Model Training: Comparing Logistic Regression, Random Forest, and XGBoost.
• Hyperparameter Tuning: Optimizing the Random Forest model with GridSearchCV.
• Evaluation: Reporting accuracy, precision, recall, and F1-score.

The final tuned model provides a robust prediction of survival based on passenger data.

---

Dataset

The dataset used is tested.csv (Titanic dataset), which includes features like:

• Pclass: Passenger class
• Sex: Gender
• Age: Age of passenger
• SibSp: Number of siblings/spouses aboard
• Parch: Number of parents/children aboard
• Fare: Ticket fare
• Cabin: Cabin information
• Embarked: Port of embarkation

Target: Survived (0 = Did not survive, 1 = Survived)

---

Requirements

• Python 3.8+
• Libraries:

  pip install pandas numpy scikit-learn xgboost

Install dependencies:

pip install -r requirements.txt

---

Project Structure

Titanic-Survival-Prediction/
├── data/
│   └── tested.csv         # Titanic dataset (not included, add your own)
├── main.py               # Main script to run the pipeline
├── README.md             # Project documentation
└── requirements.txt      # Dependencies

---

How to Run

1. Clone the repository:

   git clone https://github.com/venkat-0706/Titanic-Survival-Prediction.git
   cd Titanic-Survival-Prediction

2. Install dependencies:

   pip install -r requirements.txt

3. Add dataset: Place tested.csv in the data/ folder (or update the file path in main.py).

4. Run the script:

   python main.py

The script will:

• Load and preprocess the data
• Engineer features
• Train and evaluate models
• Tune the best model and output the final accuracy

---

Code Overview

1. Data Loading

Loads the dataset and displays initial insights (head, info, missing values).

2. Preprocessing

• Fills missing Age with median, Embarked with mode.
• Converts Cabin to binary (known/unknown).
• Encodes Sex (male: 0, female: 1) and Embarked (one-hot).
• Normalizes Age and Fare.

3. Feature Engineering

• FamilySize: Combines SibSp + Parch + 1.
• IsAlone: Flags passengers traveling alone.

4. Model Training

Trains and evaluates:

• Logistic Regression
• Random Forest
• XGBoost

Metrics: Accuracy, Precision, Recall, F1-Score.

5. Hyperparameter Tuning

Optimizes Random Forest with GridSearchCV using:

• n_estimators: [100, 200]
• max_depth: [10, 20, None]
• min_samples_split: [2, 5]

---

Results

Sample output:

Logistic Regression:
Accuracy: 0.7857
Precision: 0.7500
Recall: 0.6667
F1-Score: 0.7059

Random Forest:
Accuracy: 0.8214
Precision: 0.8000
Recall: 0.7273
F1-Score: 0.7619

XGBoost:
Accuracy: 0.8036
Precision: 0.7778
Recall: 0.7000
F1-Score: 0.7368

Best Params: {'max_depth': 10, 'min_samples_split': 5, 'n_estimators': 200}
Best CV Score: 0.8345
Final Test Accuracy: 0.8393

Note: Results may vary based on your dataset split.

---

Future Improvements

• Add visualization (e.g., confusion matrix, feature importance).
• Implement cross-validation for more robust evaluation.
• Experiment with additional features or models (e.g., SVM, Neural Networks).

---

Contributing

Feel free to fork this repo, submit issues, or send pull requests. All contributions are welcome!

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Author: Chandu Abbireddy
GitHub: github.com/Abbireddy Venkata Chandu
LinkedIn: linkedin.com/in/Abbireddy Venkata Chandu