Titanic Survival Prediction

This project implements a machine learning pipeline to predict the survival of passengers on the Titanic using various classification algorithms. The project involves data preprocessing, feature engineering, model training, hyperparameter tuning, and model evaluation.

Project Structure

• data/ - Directory containing the dataset (train.csv and test.csv).
• python_scikit-learn_titanic.py - Main script containing the code for loading data, preprocessing, training models, and generating submissions.

Running the Code

Feature Engineering & Preprocessing:

Feature engineering is applied to create new features such as Family_Size, Is_Alone, Title, Age_Group, etc. Preprocessing pipelines are defined for numerical and categorical features.

Model Training & Hyperparameter Tuning:

Several classifiers are trained and tuned using GridSearchCV, including:

• Random Forest
• Extra Trees
• XGBoost
• Decision Tree
• Logistic Regression
• Gaussian Naive Bayes
• K-Nearest Neighbors

Best models and their parameters are saved as .pkl files.

Key Libraries Used

• pandas - Data manipulation and analysis.
• numpy - Numerical computations.
• matplotlib & seaborn - Data visualization.
• scikit-learn - Machine learning library for model building and evaluation.
• xgboost - Implementation of gradient boosting algorithm.

Feature Engineering

The following features are engineered:

• Family_Size - Number of family members onboard.
• Is_Alone - Binary feature indicating if the passenger was alone.
• Title - Extracted from passenger names.
• Age_Group - Binned age groups.
• Ticket_Number, Ticket_Location - Extracted from ticket information.
• Cabin_Alphabet, Cabin_Recorded - Extracted from cabin information.

Hyperparameter Tuning

Hyperparameters are tuned using GridSearchCV with cross-validation to find the best model configuration.

Feature Importance

Feature importance is plotted for the top 20 features for each model.