Building a Movie Recommendation System

Introduction

During my internship, I worked on developing a Movie Recommendation System using collaborative filtering techniques with Singular Value Decomposition (SVD). The goal was to build a system that suggests movies based on user preferences. The dataset used was MovieLens 100K, which contains user ratings for various movies.

Throughout this journey, I faced multiple challenges, including package compatibility issues, dependency errors, and computational limitations. This blog-style README outlines how I tackled each issue and completed the project step by step.

Step 1: Setting Up the Environment

To ensure an organized workspace, I created a dedicated project folder:

C:\Users\Dell\Desktop\CODSOFT_Recommendation_System

Inside this folder, I set up a virtual environment to manage dependencies efficiently:

python -m venv venv venv\Scripts\activate

Installed the required Python libraries: pip install pandas numpy matplotlib seaborn scikit-learn surprise flask

Step 2: Downloading and Loading the MovieLens Dataset

1. MovieLens 100K dataset, which contains:

• ratings.csv – User ratings for movies
• movies.csv – Movie titles and genres

2. After downloading and extracting the dataset,loaded it into a Pandas DataFrame:

ratings = pd.read_csv("data/ratings.csv") movies = pd.read_csv("data/movies.csv")

3. Merged the datasets ensured we had complete information:

data = pd.merge(ratings, movies, on="movieId") data.to_csv("data/merged_ratings.csv", index=False)

Step 3: Challenge 1 – scikit-surprise Installation Issue

I encountered the following error:

Microsoft Visual C++ 14.0 or greater is required. So I installed Microsoft C++ Build Tools from Visual Studio and restarted my system. Then, I reinstalled the package: pip install scikit-surprise

Step 4: Implementing the SVD Model

Implemented Singular Value Decomposition (SVD) for collaborative filtering:

Step 5: Challenge 2– NumPy 2.x Compatibility Issue

I encountered the following error:

A module that was compiled using NumPy 1.x cannot be run in NumPy 2.2.3. So resolved it by downgrading NumPy pip uninstall numpy pip install numpy<2

Step 6: Generating Movie Recommendations

Implemented a function to recommend movies based on predicted ratings:

def get_recommendations(user_id, model, movies, num_recommendations=5): movie_ids = movies['movieId'].unique() predictions = [model.predict(user_id, movie_id) for movie_id in movie_ids] predictions.sort(key=lambda x: x.est, reverse=True)

top_movies = [movies[movies['movieId'] == pred.iid]['title'].values[0] for pred in predictions[:num_recommendations]]
return top_movies

Calling this function with a sample user ID:

user_id = 1 recommended_movies = get_recommendations(user_id, model, movies) print(f"Recommended Movies for User {user_id}: {recommended_movies}")

This successfully generated personalized recommendations for a given user.

Step 7: Deploying the Model with Flask

To make the recommendation system accessible as an API, I built a Flask application:

from flask import Flask, request, jsonify from collaborative_filtering import model, get_recommendations import pandas as pd

movies = pd.read_csv("data/movies.csv") app = Flask(name)

@app.route("/recommend", methods=["GET"]) def recommend(): user_id = int(request.args.get("user_id")) recommendations = get_recommendations(user_id, model, movies) return jsonify({"User ID": user_id, "Recommendations": recommendations})

if name == "main": app.run(debug=True)

python flask_app.py

By accessing /127.0.0.1:5000/recommend?user_id=1, returns JSON recommendations for the user.

Step 8: Deploying on Render

Since deploying on a personal laptop is not ideal, I hosted the Flask API on Render. Steps:

1. Create a GitHub Repository

git init git add . git commit -m "Initial commit" git remote add origin https://github.com/yourusername/repo.git git push -u origin main

2. Deploy on Render

• Go to Render
• Create a new Web Service
• Connect it to the GitHub repository
• Set the Start Command: gunicorn flask_app:app
• Deploy

Conclusion

This project was a great learning experience. Some key takeaways:

• Handling dependencies: Fixing NumPy and Visual C++ build issues.
• Understanding recommendation models: Implementing SVD-based collaborative filtering.
• Building a Flask API for easy deployment.
• Deploying on the cloud using Render.

The final system provides personalized movie recommendations and is accessible via an API. This experience improved my problem-solving skills and deepened my understanding of recommendation systems.