🎬 CineMatch

CineMatch is an interactive movie recommendation app built with Streamlit and Python, powered by the MovieLens datasets. Search a movie, view details, and explore similar titles computed from user–item ratings via cosine similarity (item–item). A user-based kNN (Pearson) path is included for future work.

---

Live Demo

https://cinematch-movie-finding.streamlit.app/
First load downloads a small dataset and builds a similarity cache.

---

App Previews

Homepage

---

Features

• Real-time search - case-insensitive, year-aware (“Toy Story (1995)” works).
• Item–item recommendations - cosine over binarized ratings, per-item Top-K.
• Auto bootstrap - fetches MovieLens 100K (default) or 1M (optional).
• Movie panel - title, genres, year, avg rating, #ratings.
• Interactive UI - click recommended titles to drill down.
• Caching - precomputed sparse Top-K similarity matrix on disk.

Note: MovieLens 32M is intentionally disabled to keep startup fast and Cloud-friendly.

---

Project Structure

CineMatch/
├─ dashboard/
│  └─ app.py
├─ recsys/
│  ├─ __init__.py
│  └─ recsys_core.py
├── requirements.txt
├── README.md        
├── LICENSE  
└─ streamlit_app.py

---

Core Libraries

Library	Purpose
streamlit	UI and interactivity
pandas	Data loading/manipulation
numpy	Numerical ops
scipy	Sparse matrices + cosine
requests	Dataset download
dataclasses	Data bundle struct
zipfile/io/os/re/json	I/O, parsing, metadata

---

Local Setup

1) Clone

git clone https://github.com/zareenrahman/CineMatch.git
cd CineMatch

2) (Recommended) Virtual env

python -m venv .venv
# Windows
.venv\Scripts\activate
# macOS/Linux
# source .venv/bin/activate

3) Install deps

pip install -r requirements.txt

4) Run the app

streamlit run dashboard/app.py

Then open in browser: http://localhost:8501

---

Deploy on Streamlit Cloud

• Push this repo to GitHub.
• On Streamlit Cloud → New app.

Set:

• Repo: zareenrahman/CineMatch
• Branch: main
• Main file: streamlit_app.py
• (Optional) Python version from runtime.txt (e.g., python-3.11.9).
• Deploy.

Why streamlit_app.py? It guarantees the correct import path for dashboard/app.py and avoids package path issues.

---

Data Source

Movie data is automatically fetched from GroupLens MovieLens Datasets:

• ml-100k or ml-1m (auto-detected)
• Used for non-commercial academic and demo purposes.

---

Quick Start (for reviewers / HR)

1. Click the Live Demo link
2. Try searching for “Toy Story” or “Matrix”
3. Click a movie → see similar recommendations
4. Explore, enjoy, and see your recommender in action

---