Individual Project Rulebook

Your journey to mastery through data science

INDIVIDUAL PROJECT RULEBOOK

Advanced Programming · Fall 2025 · HEC Lausanne

Your Final Project = 100% of Your Grade Build a machine learning or data analysis project that demonstrates your programming and analytical skills.

🎯 Project Philosophy

This is your opportunity to apply what you’ve learned to a real problem. Most projects will focus on machine learning and data analysis - comparing models, analyzing datasets, and presenting findings.

Core Principles:

  • Working Code First: If we can’t run python main.py, we can’t grade it
  • Reproducibility: Same code → same results, every time
  • Clear Documentation: README explains setup, REPORT explains findings
  • Clean Structure: Organized code that others can understand

📊 Project Focus: Machine Learning

Most projects should follow this pattern:

Research Question → Data → Models → Evaluation → Report

Example: “Which model best predicts customer churn: Random Forest, Logistic Regression, or XGBoost?”

  • Classification: Predict categories (spam detection, sentiment analysis, disease diagnosis)
  • Regression: Predict values (house prices, sales forecasting, stock returns)
  • Clustering: Find patterns (customer segmentation, anomaly detection)
  • Comparison Studies: Compare 3+ models on the same task

Custom Projects

Want to do something different? As long as you can demonstrate:

  • Data manipulation with pandas/NumPy
  • Some form of analysis or modeling
  • Clear methodology and results
  • Reproducible code

…then propose it!

📦 Deliverables

1. Project Proposal (Due: November 3)

  • Length: 300-500 words
  • Format: PROPOSAL.md in your repository
  • Content: What you’ll build, what data you’ll use, what question you’ll answer

2. GitHub Repository (Due: December 21)

Required Structure:

your-project/
├── README.md              # Setup and usage instructions
├── PROPOSAL.md            # Your project proposal
├── environment.yml        # Conda dependencies
├── requirements.txt       # Pip dependencies (alternative)
├── main.py               # Entry point - THIS MUST WORK
├── src/                  # Source code modules
│   ├── __init__.py
│   ├── data_loader.py    # Data loading and preprocessing
│   ├── models.py         # Model definitions
│   └── evaluation.py     # Evaluation and visualization
├── data/
│   └── raw/              # Original data (or instructions to download)
├── results/              # Output figures and metrics
└── notebooks/            # Jupyter notebooks (optional, for exploration)

Key Requirements:

  • python main.py must run without errors
  • All dependencies listed in environment.yml or requirements.txt
  • Data included or clear download instructions
  • Use random_state for reproducibility

3. Technical Report (Due: December 21)

  • Length: ~10 pages (excluding code appendix)
  • Format: PDF (use provided Markdown or LaTeX template)
  • Structure:
    1. Abstract (200 words): Summary of project
    2. Introduction: Research question and motivation
    3. Literature Review: Brief context and related work
    4. Methodology: Data, preprocessing, models, evaluation metrics
    5. Results: Tables, figures, findings
    6. Discussion: Interpretation, limitations
    7. Conclusion: Summary and future work
    8. References: Cited sources

Download templates from the course website:

  • Markdown template → PDF with pandoc
  • LaTeX template → PDF with pdflatex

4. Presentation (December 16-20, Optional)

  • Duration: 8-10 minutes + Q&A
  • Content: Problem, approach, demo, results, learnings

🔧 Technical Requirements

Must Have

Component Requirement
Language Python 3.10+
Entry Point python main.py runs successfully
Dependencies Listed in environment.yml or requirements.txt
Reproducibility random_state set everywhere
Documentation README with setup instructions
Report ~10 page PDF with methodology and results

Nice to Have (But Not Required)

  • Unit tests
  • CI/CD pipeline
  • Docker containerization
  • Type hints

  • 70% test coverage

Focus on making your project work well, not on adding features you don’t need.

Code Quality

  • Formatting: Use consistent style (black/autopep8 recommended)
  • Naming: Clear, descriptive variable and function names
  • Structure: Modular code in src/ directory
  • Comments: Explain why, not what

📅 Timeline

Date Milestone
Nov 3 Project Proposal Due
Nov 10 Proposal Feedback
Nov 24 Feature Freeze - stop adding, start polishing
Dec 8 Documentation Week - finalize README and report
Dec 21 Final Submission
Dec 16-20 Presentations (optional)

Development Strategy

  1. Week 1-2: Set up structure, get data, basic pipeline working
  2. Week 3-4: Implement models, run experiments
  3. Week 5-6: Analyze results, create visualizations
  4. Week 7: Write report, polish documentation

📊 What We’re Looking For

The #1 Grading Criterion

Does python main.py work on our machines?

If we can’t run your code, we can’t grade it. Test on a fresh environment before submitting!

Evaluation Focus

  • Working Code: Runs without errors, produces output
  • Clear Methodology: We understand what you did and why
  • Meaningful Results: You actually answered your research question
  • Professional Presentation: Clean code, clear report, good documentation
  • Reproducibility: We get the same results when we run it

🤖 AI Tools Policy

We encourage using AI tools (ChatGPT, Copilot, Claude) as learning aids.

✅ Good uses:

  • Debugging help
  • Learning new libraries
  • Code review suggestions
  • Documentation writing

❌ Not acceptable:

  • Submitting code you don’t understand
  • Having AI write your entire project

Required: Create AI_USAGE.md documenting how you used AI tools.

❓ Common Questions

Q: Can I use a dataset from Kaggle? A: Yes! Just cite your source and include download instructions.

Q: Do I need to write tests? A: Not required, but encouraged if you have time.

Q: What if my model doesn’t perform well? A: That’s fine! Explain why in your discussion section. Negative results are still results.

Q: Can I work with a partner? A: No, all projects are individual work.

Q: What if I can’t finish everything? A: Focus on having a working main.py and a complete report. A smaller project done well beats an ambitious project that doesn’t run.

🚀 Getting Started

  1. Fork the example project from the course repository
  2. Replace the Iris example with your own data and research question
  3. Follow the same structure - it works!
  4. Test on a fresh environment before submitting

See the examples/iris-classification/ folder in the course repository for a complete working example.

📧 Need Help?

  • Discord: Quick questions and peer help
  • Office Hours: In-depth project discussions
  • Email: Project-specific issues

We want you to succeed! Start early, ask questions, and make something you’re proud of.

*Last updated: November 2025 Deadline: December 21, 2025*