Movie Recommendation System

Project

• Description: A full-stack movie recommendation system that provides movie search and recommendations. The repository contains a React client, an Express server, and a small Python FastAPI service that serves TF-IDF based recommendations using precomputed pickled models.

Features

• Search movies: Query movie details via the OMDb API (server uses OMDB_API_KEY).
• Content-based recommendations: A Python FastAPI service returns TF-IDF recommendations from precomputed data by natural language processing.
• Simple REST API: The server exposes endpoints used by the React client.

Tech Stack

• Frontend: React (create-react-app)
• Backend: Node.js + Express
• ML service: Python + FastAPI (served by uvicorn)

Repository Structure (important files)

• client/ — React app (run with npm start)
• server/ — Node/Express API (run with npm start)
• server/python-ml/ — FastAPI TF-IDF recommender (main.py and pickled model files)

Prerequisites

• Node.js (16+) and npm
• Python 3.8+ and pip
• An OMDb API key (set as OMDB_API_KEY in server .env)

Setup & Run (Windows)

1. Clone the repo

git clone <repo-url>
cd movie-recommendation-system

2. Server (Node/Express)

• Install dependencies and set environment variables:

cd server
npm install
# create a .env file with at least:
# OMDB_API_KEY=your_key_here

• Start the server:

npm start

The server uses server.js and exposes the API used by the client. The OMDB_API_KEY environment variable is read as process.env.OMDB_API_KEY.

3. Python ML service (TF-IDF recommender)

• Create and activate a virtual environment, then install the minimal packages:

cd server/python-ml
python -m venv venv
venv\Scripts\activate
pip install fastapi uvicorn pandas numpy

• Start the FastAPI service (binds to port 8001 by default):

uvicorn main:app --host 0.0.0.0 --port 8001

Notes:

• The python-ml folder already contains df.pkl, indices.pkl, and tfidf_matrix.pkl used by main.py.
• The server calls the recommender at http://localhost:8001/recommend/tfidf (see server/utils/pythonML.js).

4. Client (React)

cd ../../client
npm install
npm start

The React app will open in the browser (usually at http://localhost:3000) and talk to the Express server for search and recommendations.

Environment variables

• Create server/.env with:

OMDB_API_KEY=your_omdb_key
# (other env vars if added later)

Run order

• Start the Python ML service (uvicorn) first (so recommender endpoint is available).
• Start the Node server next (npm start).
• Start the React client last (npm start in client/).

Testing

• Use the browser UI to search for titles and request recommendations.
• You can call recommender directly:

curl "http://localhost:8001/recommend/tfidf?title=The%20Matrix&top_n=5"

Troubleshooting

• If the client cannot fetch data, confirm the Express server is running and that CORS is enabled.
• If the recommender returns 404 for a title, the title may not exist in the pickled dataset (ensure titles are normalized and present in indices.pkl).
• If Python packages fail to install, confirm pip and Python versions; use python -m pip install --upgrade pip.