Data Science Projects

This repository is hosted at https://github.com/MediaJohnD/data-science-projects.

Clone the repository:

git clone https://github.com/MediaJohnD/data-science-projects.git
cd data-science-projects

Local development

Create a virtual environment and install dependencies:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Run lint and tests:

flake8 src tests
pytest -q

Prefect Orchestration

This project implements a small demonstration of the OptiReveal workflow using Prefect for orchestration. A minimal sample data set is bundled with the code so that the full pipeline can be executed without external dependencies. The flow consists of the following stages:

1. Ingest – load raw visit events.
2. Feature Engineering – aggregate visits into device level features.
3. Model Training – train a logistic regression model predicting visitor propensity.
4. Monitoring – log basic metrics.
5. Triggers – create simple contextual triggers.
6. Deployment – expose a FastAPI scoring service.

Run the end-to-end flow locally:

python src/pipeline.py

📓 Notebook Walkthrough

An end-to-end example of the pipeline can be found in full_pipeline_summary.ipynb.