SpeedSense 🌐

SpeedSense is the lightweight, public-facing frontend for our AI-powered 5G throughput prediction system — designed for telecom operators and researchers.
This interface connects to our full backend stack to deliver real-time predictions, explainable insights, and PDF exports.
🌍 Live site: speedsense.me
🤖 Full ML/MLOps Project: QosMLOPS GitHub Repo

---

This work was developed as part of the Integrated Project at Esprit School of Engineering, under the guidance of professors Rahma Bouraoui, Safa Cherif, and Zaineb Labidi.

---

🖼️ Preview

More in the assets/ folder.

---

✨ What You Can Do with SpeedSense

• 📥 Upload network performance data in CSV format
• 📈 Get real-time 5G throughput predictions powered by XGBoost
• 🧠 Understand model behavior with SHAP explainability
• 💬 Generate GPT-4o-powered QoS summaries for every prediction
• 📄 Export results as PDF
• 🖐️ Or use our manual input form (for users without a CSV file)

---

🤖 Chatbot (LLM FAQ Assistant)

SpeedSense includes a smart FAQ chatbot trained on 30+ domain-specific questions.
Built using Gemini + vector search (RAG)

This chatbot is embedded in the About Us page.

---

🧠 AI Insight System

Each prediction includes:

• SHAP values showing feature impact
• GPT-4o-generated insights:
  • QoS Rating
  • Key influencing features
  • Optimization tips
• Displayed directly in-browser
• Exportable as PDF

---

📁 Folder Structure

.
├── app.py                    # Main FastAPI app
├── Dockerfile                # Docker container setup
├── Makefile                  # CLI commands (build, run, etc.)
├── README.md                 # Project overview
├── requirements.txt          # Python dependencies
├── .env                      # Local environment variables
├── .gitignore
│
├── assets/                   # Images and demo screenshots
│   └── ... (e.g. screen1.png, gif.gif)
│
├── Models/                   # Trained ML model files (e.g. best_xgb_model.pkl)
│
├── services/                 # All backend logic
│   ├── __init__.py
│   ├── chatgpt_service.py        # GPT-powered explanation generator
│   ├── embedding_utils.py        # Gemini embeddings wrapper
│   ├── faq_data.py               # FAQ base content (optional after caching)
│   ├── faq_index.pkl             # Precomputed FAQ vector index
│   ├── faq_search.py             # RAG search logic (cosine similarity)
│   ├── generate_faq_index.py     # Script to create faq_index.pkl
│   └── preprocessing.py          # Data transformation for prediction
│
├── frontend/                # Public site UI
│   ├── aboutus.html              # Main About Us page
│   ├── csv.html                  # CSV upload input page
│   ├── explain.html              # SHAP explanation page
│   ├── index.html                # Homepage
│   ├── predict.html              # Manual input form
│   │
│   ├── components/               # Reusable page components
│   │   └── chatbot.html              # LLM FAQ assistant (auto-included in About Us)
│   │
│   ├── css/                      # Stylesheets (global + chatbot-specific)
│   ├── js/                       # JavaScript logic (if decoupled from HTML)
│   ├── images/                   # Icons, logos, illustrations
│   └── webfonts/                 # Font files

---

⚙️ Local Setup

git clone https://github.com/RideneFiras/SpeedSense_Front.git
cd SpeedSense_Front
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn app:app --reload

Then open http://127.0.0.1:8000

---

📬 Contact

For feedback or support: contact@speedsense.me