An AI-powered web application that helps content creators and casual viewers understand what’s trending on YouTube and predict how videos will perform — all in one place.
It combines machine learning predictions,real-time YouTube Data API analytics, and natural language summarization to save hours of manual research.
- 30-Day View Prediction – Predict future views of any YouTube video using a trained ML regression model.
- AI Summaries – Instantly summarize trending videos so you can grasp the content without watching it.
- Category & Filter Search – Browse trending videos by category, with an option to exclude Shorts.
- Real-time Analytics – Fetch live statistics (views, likes, comments, etc.) directly from YouTube.
- Content Creators's Help – Understand why videos trend and plan content strategically.
- Frontend: HTML5, CSS3, JavaScript
- Backend: Flask (Python)
- Machine Learning: Scikit-learn, Joblib, Pandas, NumPy
- NLP: Hugging Face Transformers, NLTK
- API: YouTube Data API v3 (Google API Python Client)
- Models:
- sshleifer/distilbart-cnn-12-6 (fine-tuned for summarization)
- Gradient Boosting Regressor (view prediction)
-
To Analyze a Video: Paste the YouTube video URL → Fetch stats → Get AI summary + 30-day view prediction.
-
To Browse Trending Content: Select category → View summaries + performance estimates for top trending videos.
- Summarization: Fine-tuned DistilBART model evaluated using ROUGE scores.
- Prediction Model: Gradient Boosting Regressor with R² = 0.7984 (explains ~80% of variation).