How anurag2808/drug-2-drug-interaction works
Project type and purpose
• From the README (“Drug-2-Drug-Interaction: A ML project that predicts the drug–drug interaction”) and the file tree, we see:
• A Jupyter notebook (main.ipynb) for exploration and model development
• A CSV data file (Id_details_new.csv)
• A Python application script (app_new.py)
• Two HTML files (index.html, Details.html) for the UI
• This is a small full-stack ML application: a model training/experimentation component plus a web service for model inference exposed via a simple HTML-based frontend.
File structure
• Top-level files only; no nested directories
• main.ipynb: experiment & training pipeline
• Id_details_new.csv: raw or processed dataset
• app_new.py: backend web service (likely Flask or FastAPI)
• index.html & Details.html: frontend pages for user interaction
• Details.html may show prediction outputs or drug information
• No external database—data stored in CSV and model artifact likely saved to disk within app_new.py or loaded in memory.
README for additional insights
• Very brief; confirms ML prediction service
• No mention of external services or cloud deployments
• Implies single-node, on-premise or local web app
Elements
– Labels: “Browser UI,” “Frontend (HTML)”, “Backend API (app_new.py)”, “Data Store (CSV & model file)”, “Training Pipeline (main.ipynb)”
• From Browser to Frontend (user request)
• Frontend → Backend (HTTP POST/GET)
• Backend → Data Store (file read)
• Backend → Frontend (prediction response)
• In development view: CSV → Notebook → Model artifact → Backend
• Blue rectangles for compute components (Notebook, API)
Detailed architectural capture
• Don’t just show “ML service”—break it into “Feature Extraction,” “Model Loading,” “Prediction Engine” if needed
• Explicitly show where preprocessing happens (notebook vs. API)