🚗 Traffic Analysis: Palaiseau ↔ Paris

A Streamlit app analyzing commute time data (collected from a GPS API) between Palaiseau and Paris, used to find the best time of day to travel.

Project structure

traffic_analysis/
├── app.py                  # Streamlit UI
├── src/
│   └── data_loader.py       # Data loading/filtering logic (unit tested)
├── tests/
│   └── test_data_loader.py  # Pytest suite
├── data/
│   ├── data_X_solfe.json    # Palaiseau -> Paris trip records
│   └── data_solfe_X.json    # Paris -> Palaiseau trip records
├── Dockerfile
├── requirements.txt
├── requirements-dev.txt
└── .github/workflows/ci.yml # CI: lint, test, Docker build

Data format

Each JSON file is a list of trip records:

{
  "timestamp": "2025-10-05T15:12:52.076311",
  "duration_seconds": 2319,
  "duration_minutes": 38.65,
  "distance_meters": 26041,
  "distance_km": 26.041,
  "base_duration": 1921,
  "traffic_delay": 398
}

Running locally

python -m venv venv
source venv/bin/activate
pip install -r requirements-dev.txt
streamlit run app.py

Running tests

pytest tests/ -v --cov=src

Running with Docker

docker build -t traffic-analysis .
docker run -p 8501:8501 traffic-analysis

Then open http://localhost:8501.

CI

GitHub Actions (.github/workflows/ci.yml) runs on every push/PR to main:

1. Lints src/ and app.py with flake8
2. Runs the pytest suite (with coverage) on Python 3.11 and 3.12
3. Builds the Docker image to confirm it's buildable

Reproducibility notes

• Dependencies are pinned in requirements.txt / requirements-dev.txt.
• Data loading and transformation logic lives in src/data_loader.py, separate from the Streamlit UI, so it can be unit tested without a browser.
• The Docker image installs dependencies before copying code, so rebuilds are fast when only app logic changes.