How ali-islam111/heart-disease-risk-prediction works
Project type and purpose
• This is a full-stack machine-learning web application for heart-disease risk prediction.
• The pipeline covers data ingestion, preprocessing, feature engineering, modeling (both supervised and unsupervised), hyperparameter tuning, model export, and a Streamlit-based UI exposed via an ngrok tunnel.
File structure
• Top-level folders:
• UI: contains app.py (Streamlit frontend)
• model: contains the serialized pipeline final_tuned_pipeline.pkl
• datasets: raw and intermediate CSVs for preprocessing, feature selection, PCA
• notebooks: exploratory and modeling steps (preprocessing, EDA, feature selection, PCA, supervised & unsupervised learning, hyperparameter tuning)
• deployment: the script deploy_with_ngrok.py that launches Streamlit and opens an ngrok tunnel
• results: evaluation output (Results.txt)
• Config files: requirements.txt for dependencies, README.md for project overview.
README for additional insights
• The README outlines project phases, technology stack (Python, Streamlit, scikit-learn, ngrok), and how the final artifact (final_tuned_pipeline.pkl) is consumed by app.py.
• Deployment instructions show how the pieces fit at runtime: the UI loads the model, exposes endpoints, and streams results to users via an ngrok URL.
Elements
• Blue rectangles for data pipelines
• Orange rectangles for runtime services (Streamlit, ngrok)
Detail without overcomplicating
• Every stage: raw data → preprocessing → feature selection → PCA → model training → model export → UI → deployment.
• Notebooks block as “Development / Exploration” but do not detail every notebook—group by phase.
• That deploy_with_ngrok.py orchestrates both launching Streamlit and creating the ngrok tunnel.