KWESI-LAWER/titanic-data-science-project

Titanic survival prediction and exploratory data analysis project for the BCS 404 Data Science course at Accra Technical University. # BCS 404: Titanic Data Science Project **Accra Technical University — Department of Computer Science** **Course:** BCS 404: Introduction to Data Science with Python **Lecturer:** Dr. Joseph Dadzie **Academic Year:** 2025/2026, Second Semester ## Project Overview This project performs a complete data science workflow on the Kaggle Titanic dataset: data acquisition, data cleaning, exploratory data visualisation, statistical analysis, and a Logistic Regression machine learning model to predict passenger survival. ## Repository Contents | File | Description | |---|---| | `Titanic_Project_BCS404.ipynb` | Jupyter Notebook containing all code, outputs, and visualisations for Tasks 1–6 | | `Titanic_Project_Report_BCS404.docx` | Full project report (cover page, TOC, introduction through appendix) | | `train.csv` | Titanic training dataset (from Kaggle) | | `titanic_clean.csv` | Cleaned dataset produced by the notebook | | `README.md` | This file | ## How to Run 1. Install dependencies: `pip install pandas numpy matplotlib seaborn scikit-learn jupyter` 2. Place `train.csv` in the same directory as the notebook. 3. Launch Jupyter: `jupyter notebook Titanic_Project_BCS404.ipynb` 4. Run all cells in order. ## Summary of Results - **Dataset:** 891 passengers, 12 columns (Kaggle Titanic training set). - **Cleaning:** `Age` imputed by Pclass/Sex median; `Embarked` imputed by mode; `Cabin` replaced with a `HasCabin` indicator due to 77% missingness. - **Key correlations:** Strongest positive — Fare & HasCabin (r = 0.482). Strongest negative — Pclass & HasCabin (r = -0.726). - **Model:** Logistic Regression on 8 features, 80/20 stratified train/test split. - **Accuracy:** 81.0% on the test set (179 passengers). ## Dataset Source Kaggle "Titanic - Machine Learning from Disaster" competition: kaggle.com ## Author [Insert Your Name] — [Insert Your Index Number]