Titanic ML Lab (Beginner Friendly)

This project trains a RandomForestClassifier on the Titanic dataset and creates a submission.csv file for test predictions.

Project Structure

• src/preprocess.py: data loading, cleaning, feature engineering, and preprocessing pipeline
• src/train.py: model training and validation accuracy report
• src/predict.py: test-set prediction and submission.csv creation
• notebooks/titanic_analysis.ipynb: beginner walkthrough notebook
• requirements.txt: required Python packages

Requirements

• Python 3.9+
• Titanic files:
  • data/train.csv
  • data/test.csv

Install dependencies:

pip install -r requirements.txt

What This Project Does

1. Reads train.csv and test.csv from data/
2. Performs data cleaning:
   • fills missing Age and Fare with median
   • fills missing Embarked with most frequent value
3. One-hot encodes categorical columns (Sex, Embarked)
4. Creates new features:
   • FamilySize = SibSp + Parch + 1
   • IsAlone (1 if family size is 1, else 0)
5. Trains a RandomForestClassifier
6. Evaluates accuracy on a validation split
7. Generates submission.csv for test predictions

Run

From project root:

python src/train.py
python src/predict.py

After running, you should see:

• models/model.joblib
• submission.csv

Pipeline Flow

Training Flow (python src/train.py)

flowchart TD
    A["train main"] --> B["load data"]
    B --> C["read train csv and test csv"]
    C --> D["prepare train features"]
    D --> E["add family features"]
    E --> F["prepare X and y"]
    F --> G["train test split"]
    G --> H["build preprocessor"]
    H --> I["fit transform train"]
    I --> J["transform validation"]
    J --> K["fit random forest"]
    K --> L["predict validation"]
    L --> M["compute accuracy"]
    M --> N["save model with joblib"]

Loading

Prediction Flow (python src/predict.py)

flowchart TD
    A["predict main"] --> B{"model joblib exists"}
    B -- "no" --> C["raise file not found"]
    B -- "yes" --> D["load data"]
    D --> E["select test dataframe"]
    E --> F["load model and preprocessor"]
    F --> G["prepare test features"]
    G --> H["add family features"]
    H --> I["transform test features"]
    I --> J["predict survived"]
    J --> K["build submission dataframe"]
    K --> L["write submission csv"]

Loading

Notes for Beginners

• Start with notebooks/titanic_analysis.ipynb if you want to understand each step interactively.
• The scripts in src/ are the same logic in reusable Python modules.