CineBlend: The Mathematical Movie Matchmaker | Python, Scikit-Learn & Streamlit
CineBlend: The Mathematical Movie Matchmaker | Machine Learning Project (Python & Streamlit)
🚀 *Project Overview*
This is a Machine Learning web application that goes beyond simple movie suggestions by using Vector Space Modeling to find mathematical relationships between films. Built with Python and Streamlit, the app features a "Dual Mode" engine that allows users to not only find similar movies but also "breed" two completely different films to find the hidden gems that bridge the gap between them.
🔍 *Key Features Demonstrated:*
*Dual Mode Architecture:* Seamless switching between "Classic Recommend" and "The Blend" modes.
*Vector Arithmetic:* A unique algorithm that calculates the "Mean Vector" (Mathematical Midpoint) of two movies to generate a hybrid recommendation.
*Content-Based Filtering:* NLP analysis of Movie Genres, Keywords, Cast, and Crew to build 5000-dimensional feature vectors.
*Premium UI Design:* Integrated `streamlit-shadcn-ui` for a modern, card-based aesthetic that mimics professional SaaS products.
*Local Computation:* A privacy-first approach running 100% locally with no external API dependencies.
🛠️ *Tech Stack:*
*Language:* Python 3.14
*Frontend:* Streamlit (with Shadcn UI Components)
*ML Core:* Scikit-Learn (CountVectorizer, Cosine Similarity)
*Data Processing:* Pandas, NumPy
🔗 *Links:*
*GitHub Repository:* https://github.com/Dhruv-Mann/CineBle...
*Dataset:* TMDB 5000 Movies (Kaggle)
⏱️ *Timestamps:*
0:00 - Project & Code Structure
0:08 - Launching the App
0:12 - Mode A: Classic Recommendation (Avatar)
0:22 - Mode B: The Blend Logic (Dual Mode)
0:33 - Mathematical "Child" of Two Movies