How srotriyosengupta/creditcardfrauddetectionpipeline works
Project type and purpose
• This is a production-grade, automated machine-learning pipeline for credit-card fraud detection.
• It uses Airflow DAGs to orchestrate model training, shadow-model generation, and prediction validation.
• CI/CD is configured via GitHub Actions; the entire pipeline can be containerized with Docker.
File structure
• Top-level:
• dags/ — contains Airflow DAG definitions (fraudvalidation_dag.py)
• scripts/ — Python scripts for training (train_model.py), shadow-model generation, and validation
• models/ — serialized model artifacts (prod_model.pkl, shadow_model.pkl)
• tests/ — unit tests (test_validation_logic.py)
• .github/workflows/ci.yml — CI/CD pipeline definition
• dockerfile — container build instructions
• requirements.txt — Python dependencies
• Patterns:
• Orchestration layer (Airflow) triggers individual script components
• Clear separation: compute (scripts), storage (models/), orchestration (dags/), CI/CD (.github)
README for additional insights
• Confirms use of IBM’s synthetic dataset.
• Emphasizes end-to-end pipeline: training → shadow validation → testing → deployment checks.