How ambarish-224/end-to-end-insurance-premium-prediction works
Project type and purpose
• This is an end-to-end Machine Learning application for Insurance Premium Prediction. It consists of:
• A modular ML training pipeline (data ingestion → validation → transformation → training → evaluation → model pusher)
• A file-based artifact and model registry
• A lightweight inference web UI (Streamlit)
• Containerization (Docker) and CI/CD (GitHub Actions) for automated training and deployment
File structure
• Top-level:
• Dataset/insurance.csv – raw data source
• Insurance/ – Python package for all pipeline components, config, entities, exceptions, logger, utils
– Insurance/components – data_ingestion.py, data_validation.py, data_transformation.py, model_trainer.py, model_evaluation.py, model_pusher.py
– Insurance/pipeline – training_pipeline.py and batch_prediction.py orchestrators
– Insurance/config.py, entity/, exception/, logger/, utils.py, predictor.py
• artifact/ – timestamped folders holding intermediate files (feature_store, datasets, transformed arrays, reports, pickled models)
• saved_models/ – model registry for deployment (model.pkl, transformer.pkl, target_encoder.pkl)
• streamlit_app.py, main.py, demo.py – entrypoints for serving and testing
• configs/config.yaml – pipeline configuration
• Dockerfile, .github/workflows/main.yaml – container and CI/CD setup
README for additional insights
• Confirms a six-stage ML pipeline
• Mentions technologies: Python 3.7, Docker, Streamlit, CI/CD