🎥 YouTube Analytics Dashboard

A machine learning-based dashboard to predict YouTube video engagement metrics (likes, views) using initial metadata.

---

📂 Folder Structure

.
├── app.py                     # Flask backend (prediction API)
├── dashboard.py              # Streamlit frontend
├── models/                   # Trained model pickles
│   ├── LGBM_Blend_Model.pkl
│   ├── XGB_Blend_Model.pkl
│   ├── CatBoost_Blend_Model.pkl
│   └── Blending_with_Lasso.pkl
├── data/
│   └── preprocessed_youtube_data.csv
├── logs/                     # Log files (created at runtime)
├── requirements.txt
└── README.md

---

🔧 Setup Instructions

1. Clone the repo

git clone <your-repo-link>
cd yt-analytics-dashboard

2. Create & activate virtual environment

python3 -m venv venv
source venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

---

🚀 Run the Application

Backend (Flask)

nohup python app.py > logs/backend.log 2>&1 &

or

python app.py

Now the back end is running at port 5000 with /predict end point.

Frontend (Streamlit)

nohup streamlit run dashboard.py --server.port=8501 --server.address=0.0.0.0 > logs/frontend.log 2>&1 &

or

streamlit run dashboard.py

Access dashboard at: http://<your-server-ip>:8501

---

📈 Model Info

Blended regression model using:

• LightGBM
• XGBoost
• CatBoost
• LASSO (meta model)

Trained on YouTube metadata from multiple categories.

---

🔍 Related GitHub Resources

📝 Training Notebook

Repo: YT-analytics

🔹 YouTube Scraper

Script: youtube_scraper.py

---

🌟 Project Contributors

GitHub Profile Pic	Name	University	GitHub Link
	Parishith Ragumar	CSE, University of Moratuwa	@ragupari
	Birunthaban Rajendram	CSE, University of Moratuwa	@birunthabanr
	Donald Aadithiyan	CSE, University of Moratuwa	@DonaldAadithiyan

---

💡 Future Ideas

• Add weekly trend predictions
• Integrate more features like tags, titles (via embeddings)
• Visualize category-wise averages and comparisons