🚢 Titanic Survival Prediction — AI Project

Predict which passengers survived the Titanic shipwreck using machine learning — a classic binary classification challenge from Kaggle.

Academic project (2025–2026) supervised by Prof. N. ABOUTABIT

---

📖 Table of Contents

• Overview
• Dataset
• Project Structure
• Features & Preprocessing
• Models & Validation
• Results
• Installation
• Usage
• Reports
• Contributing
• License
• Acknowledgements

---

🧠 Overview

This project tackles the Kaggle Titanic competition, one of the most popular introductory machine learning challenges. The goal is to build a predictive model that answers:

"What sorts of people were more likely to survive the Titanic disaster?"

The repository implements a complete ML pipeline:

• Exploratory Data Analysis (EDA)
• Missing value handling & preprocessing
• Feature engineering
• Model training + evaluation
• Prediction generation and submission export

---

📊 Dataset

The dataset is provided by Kaggle and contains information about 891 passengers (train) and 418 passengers (test).

File	Description
data/train.csv	Training data with survival labels
data/test.csv	Test data for prediction submission
data/gender_submission.csv	Sample submission file

Key Features

Feature	Description
Survived	Target variable (0 = No, 1 = Yes)
Pclass	Passenger class (1st, 2nd, 3rd)
Name	Passenger name
Sex	Gender
Age	Age in years
SibSp	# of siblings/spouses aboard
Parch	# of parents/children aboard
Ticket	Ticket number
Fare	Passenger fare
Cabin	Cabin number
Embarked	Port of embarkation (C, Q, S)

---

📁 Project Structure

This reflects the current repository layout.

Titanic_Machine_Learning/
├── data/
│   ├── train.csv
│   ├── test.csv
│   └── gender_submission.csv
├── src/
│   ├── data/              # loading / preprocessing modules
│   ├── features/          # feature engineering
│   ├── models/            # training / evaluation / optimization
│   ├── utils/             # config + helpers
│   └── visualization/     # EDA plots
├── main.py                # runs the full pipeline end-to-end
├── REPORT.md              # detailed report (markdown)
├── DELIVERABLES.md        # checklist of deliverables
├── Rapport_Project_AI.pdf # full technical report (PDF)
└── requirements.txt

---

⚙️ Features & Preprocessing

Key engineering steps used in the pipeline include:

• Missing value imputation

  • Age: imputation strategy based on passenger groups (e.g., by class/sex)
  • Embarked: mode imputation
  • Fare: median imputation (mainly affects test set)

• Feature engineering

  • FamilySize = SibSp + Parch + 1
  • IsAlone flag
  • Title extracted from Name (Mr, Mrs, Miss, Master, Rare, ...)
  • HasCabin (Cabin present vs missing)

• Encoding

  • One-hot encoding for categorical features (Sex, Embarked, Title, etc.)

• Scaling

  • StandardScaler applied before model training (see main.py)

---

🤖 Models & Validation

The project trains and evaluates classification models (baseline + optimized) and uses:

• Train/Validation split (stratified)
• Cross-validation
• Metrics: Accuracy, Precision, Recall, F1-score, ROC-AUC

For the exact implementation details and metrics, see REPORT.md.

---

📈 Results

A short summary (update with your Kaggle score/rank):

Item	Value
Best local validation accuracy	~80–82% (varies by model/settings)
ROC-AUC	~0.84–0.86
Kaggle public score	Add your score
Kaggle rank	Add your rank / Top X%

---

🛠️ Installation

Prerequisites: Python 3.9+

# 1) Clone the repository
git clone https://github.com/msabr/Titanic_Machine_Learning.git
cd Titanic_Machine_Learning

# 2) (Recommended) create a virtual environment
python -m venv venv
source venv/bin/activate       # On Windows: venv\Scripts\activate

# 3) Install dependencies
pip install -r requirements.txt

Note: requirements.txt currently lists core libraries (pandas/numpy/sklearn/matplotlib/seaborn).

---

🚀 Usage

Run the full pipeline (recommended)

python main.py

This will run the pipeline end-to-end (EDA → preprocessing → training → evaluation → predictions).

Data location

Make sure the Kaggle CSV files exist in:

data/train.csv
data/test.csv

---

📑 Reports

• REPORT.md — detailed methodology, preprocessing, evaluation, and recommendations
• Rapport_Project_AI.pdf — full technical report (PDF)
• DELIVERABLES.md — deliverables summary / checklist

---

🤝 Contributing

Contributions are welcome:

1. Fork the repository
2. Create a branch (git checkout -b feature/improvement)
3. Commit changes (git commit -m "Add improvement")
4. Push (git push origin feature/improvement)
5. Open a Pull Request

---

• Team: Mohamed SABR, Abdejlil SALMI, Soufaine ZEKAOUI, Anass LAMHADAR