Introduction

The Titanic dataset is one of the most widely used introductory datasets in data science and machine learning. Despite its age and simplicity, it presents realistic challenges: missing data, mixed data types, categorical features, non-linear relationships, and strong human-interest hypotheses ("women and children first", social class privilege, etc.).

This end-to-end project walks through the complete lifecycle of a data-science workflow:

1. Loading and exploring the dataset
2. Understanding relationships with plots
3. Formulating hypotheses
4. Testing hypotheses statistically
5. Preparing data for modelling
6. Training classical machine-learning models (scikit-learn)
7. Training a neural network (Keras)
8. Comparing model performance
9. Drawing conclusions

Jupyter Notebook

All the code sections have been combined in to a Jupyter notebook in this repository. It is only the code parts, none of the text explanations.

Open