🧠 ML Visual Explorer

Interactive hyperparameter visualization for machine learning models — no code needed.

Quick Start

You need two terminals — one for backend, one for frontend.

Terminal 1 — Backend (FastAPI + scikit-learn)

cd ml-explorer
chmod +x start_backend.sh
./start_backend.sh

Backend runs at: http://localhost:8000
API docs at: http://localhost:8000/docs

Terminal 2 — Frontend (React + Vite)

cd ml-explorer
chmod +x start_frontend.sh
./start_frontend.sh

Frontend runs at: http://localhost:3000

---

Project Structure

ml-explorer/
├── backend/
│   ├── main.py            ← FastAPI app, all ML models
│   └── requirements.txt   ← Python dependencies
├── frontend/
│   ├── src/
│   │   ├── App.jsx                         ← Main layout + navigation
│   │   ├── api/index.js                    ← API client
│   │   └── components/
│   │       ├── LinearRegression.jsx        ← Main interactive page
│   │       ├── SliderControl.jsx           ← Reusable slider
│   │       ├── MetricCard.jsx              ← R², MSE cards
│   │       └── FitStatusBadge.jsx          ← Overfitting/underfitting badge
│   ├── index.html
│   └── package.json
├── start_backend.sh
└── start_frontend.sh

---

Phase 1 — What's Built

Models

• ✅ Linear Regression
• ✅ Polynomial Regression (degree 1–10)
• ✅ Ridge Regression (L2 regularization)
• ✅ Lasso Regression (L1 regularization)
• ✅ ElasticNet (L1 + L2)

Datasets

• ✅ Linear trend
• ✅ Polynomial (quadratic)
• ✅ High noise
• ✅ Sine wave

Features

• ✅ Live graph updates on every slider change (debounced 300ms)
• ✅ Train/test split with color-coded scatter points
• ✅ R² and MSE metrics (train + test)
• ✅ Overfitting / underfitting detector badge
• ✅ Coefficient display
• ✅ Parameter tooltips explaining each hyperparameter

---

Coming in Phase 2

• Logistic Regression + decision boundaries
• KNN, SVM, Decision Tree, Random Forest
• Dataset upload (CSV)
• Side-by-side model comparison
• Bias-variance tradeoff curve