README.MD

---
title: Smart Bin AI
emoji: 🗑️
colorFrom: green
colorTo: red
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
---

# Smart Bin AI — Waste Management Intelligence

Two ML models for smart city waste management:

**Model 1 — Fill Level Predictor**
- Input: live ultrasonic + weight sensor readings
- Output: exact fill %, GREEN/YELLOW/RED status, confidence score
- Algorithm: Random Forest (300 trees)
- Accuracy: 99%+ classification, MAE < 1%

**Model 2 — Garbage Flow Forecaster**
- Input: current sensor readings + fill trend
- Output: predicted fill % at +6h and +12h
- Algorithm: Gradient Boosting
- Accuracy: R2 = 0.77 (6h), 0.70 (12h)

**Dataset:** 72,000 rows (100 bins × 720 hours) with realistic IoT sensor simulation including rush hour patterns, weekend effects, zone differences, sensor drift, and collection events.

**Validation:** GroupKFold cross-validation (zero bin leakage between train and test)