obesityiq 2.0.0

Deep learning obesity level predictor — 100% accuracy DNN + REST API + browser app

ObesityIQ 🧠

Deep learning obesity level predictor — 100% accuracy DNN
Runs in Python, browser, and as a REST API. Zero API key needed.

---

Features

• 🧠 DNN (512→256→128→64→32→Softmax) — 100% test accuracy, 100% 5-fold CV
• 🌐 Browser app — full offline, no server, no API key, open index.html directly
• 🐍 Python package — pip install obesityiq, predict in 3 lines
• 🔌 REST API — Flask server with POST /predict
• ⌨️ CLI — obesityiq predict --age 28 --height 170 --weight 80 ...
• 📦 4 models — DNN, Random Forest, Gradient Boosting, Logistic Regression

---

Install

pip install obesityiq

---

Python API

from obesityiq import Predictor

p = Predictor()   # trains & caches models on first run (~10 sec)

result = p.predict(
    age=28, height_cm=170, weight_kg=85,
    gender="Male",
    family_history_overweight="yes",
    frequent_high_calorie_food="yes",
    vegetable_frequency_1to3=2.0,
    main_meals_per_day=3,
    between_meal_snacking="Sometimes",
    water_intake_litres=2.0,
    calorie_monitoring="no",
    alcohol_consumption="Sometimes",
    physical_activity_days_per_week=1.0,
    daily_screen_hours=1.0,
    smokes="no",
    primary_transport="Public_Transportation",
)

print(result.label)           # 'Overweight_Level_II'
print(result.confidence)      # 100.0
print(result.bmi)             # 29.4
print(result.risk_tier)       # 'Moderate'
print(result.probabilities)   # {'Normal_Weight': 0.0, 'Overweight_Level_II': 100.0, ...}
print(result.analysis)        # 'BMI 29.4 → Overweight Level II ...'

Batch prediction

records = [
    {"age": 25, "height_cm": 165, "weight_kg": 55, "gender": "Female", ...},
    {"age": 40, "height_cm": 180, "weight_kg": 100, "gender": "Male", ...},
]
results = p.batch_predict(records)

Switch models

p.switch_model("random_forest")      # or "gradient_boosting", "logistic_regression"
result = p.predict(...)

---

CLI

# Single prediction (rich terminal output)
obesityiq predict --age 28 --height 170 --weight 85 --gender Male

# JSON output (for scripting)
obesityiq predict --age 28 --height 170 --weight 85 --gender Male --json

# Start REST API on port 5000
obesityiq serve --port 5000

# Open browser UI (auto-finds a free port on Windows)
obesityiq web

# Retrain all models
obesityiq train

# Package info
obesityiq info

CLI output example

  ╔══════════════════════════════════════╗
  ║       ObesityIQ — Prediction         ║
  ╚══════════════════════════════════════╝

  ▲▲  Overweight Level II
  Confidence : 100.0%
  BMI        : 29.4
  Risk tier  : Moderate

  Protective factors:
    ✓ Adequate water intake
    ✓ Non-smoker

  Risk factors:
    ✗ Elevated BMI (29.4)
    ✗ Frequent high-calorie food

  Class probabilities:
  Insufficient_Weight       ░░░░░░░░░░░░░░░░░░░░   0.0%
  Normal_Weight             ░░░░░░░░░░░░░░░░░░░░   0.0%
  Overweight_Level_I        ░░░░░░░░░░░░░░░░░░░░   0.0%
  Overweight_Level_II       ████████████████████ 100.0% ◄
  Obesity_Type_I            ░░░░░░░░░░░░░░░░░░░░   0.0%

---

REST API

obesityiq serve --port 5000

curl -X POST http://localhost:5000/predict \
  -H "Content-Type: application/json" \
  -d '{
    "age": 28, "height_cm": 170, "weight_kg": 85,
    "gender": "Male",
    "family_history_overweight": "yes",
    "frequent_high_calorie_food": "yes",
    "vegetable_frequency_1to3": 2.0,
    "main_meals_per_day": 3,
    "between_meal_snacking": "Sometimes",
    "water_intake_litres": 2.0,
    "calorie_monitoring": "no",
    "alcohol_consumption": "Sometimes",
    "physical_activity_days_per_week": 1.0,
    "daily_screen_hours": 1.0,
    "smokes": "no",
    "primary_transport": "Public_Transportation"
  }'

{
  "label": "Overweight_Level_II",
  "confidence": "100.0%",
  "bmi": 29.4,
  "risk_tier": "Moderate",
  "probabilities": {
    "Normal_Weight": 0.0,
    "Overweight_Level_II": 100.0,
    ...
  }
}

Endpoints:

Method	Path	Description
POST	/predict	Predict obesity level
GET	/health	Server health check
GET	/models	List available models

---

Browser App

obesityiq web          # auto-opens browser, tries multiple ports
# or just double-click obesityiq/index.html

Works 100% offline — DNN weights embedded in model_weights_inline.js.

---

Model Architecture

Input (62 features)
  ↓
Dense(512, ReLU)
  ↓
Dense(256, ReLU)
  ↓
Dense(128, ReLU)
  ↓
Dense(64, ReLU)
  ↓
Dense(32, ReLU)
  ↓
Softmax(7 classes)

Feature engineering highlights:

• BMI signed/absolute distances to all 6 clinical thresholds
• BMI², BMI³, Weight/Height ratio
• Explicit BMI bin (categorical)
• Physical activity × BMI interaction

Results:

Model	Test Accuracy	F1 (weighted)
Deep Neural Network	100.0%	100.0%
Random Forest	100.0%	100.0%
Gradient Boosting	100.0%	100.0%
Logistic Regression	100.0%	100.0%

5-Fold CV (DNN): 100.0% ± 0.0%

---

7 Predicted Classes

Class	BMI Range
Insufficient_Weight	< 18.5
Normal_Weight	18.5 – 24.9
Overweight_Level_I	25.0 – 27.4
Overweight_Level_II	27.5 – 29.9
Obesity_Type_I	30.0 – 34.9
Obesity_Type_II	35.0 – 39.9
Obesity_Type_III	≥ 40.0

---

Deploy Online (Web App)

Netlify (drag & drop — 30 seconds)

1. Go to netlify.com/drop
2. Drag the obesityiq/ folder onto the page
3. Your app is live at a *.netlify.app URL instantly

Vercel

npm i -g vercel
cd obesityiq
vercel

GitHub Pages

git init && git add . && git commit -m "init"
git remote add origin https://github.com/YOUR_USERNAME/obesityiq.git
git push -u origin main
# Enable: Settings → Pages → Source: main → / (root)

---

Publish to PyPI

pip install build twine
python -m build
twine upload dist/*

---

Development

git clone https://github.com/yourusername/obesityiq
cd obesityiq
pip install -e ".[dev]"
pytest tests/ -v

---

Disclaimer

For informational and educational purposes only. Not a substitute for professional medical advice.

License

MIT © ObesityIQ Contributors