# Movie Recommendation System

This project uses a dataset of movies, user ratings, and other metadata to generate movie recommendations. The recommendations are generated using a combination of TF-IDF vectorization, K-means clustering, and the OpenAI GPT-3.5 API.

## Datasets

The following datasets are used in this project:
- `genome-tags.csv`
- `genome-scores.csv`
- `ratings.csv`
- `tags.csv`
- `links.csv`
- `movies.csv`

## Prerequisites

Ensure you have the following Python packages installed:
- pandas
- scikit-learn
- requests
- openai

You can install the required packages using the following command:
```bash
pip install pandas scikit-learn requests openai

API Key

Replace YOUR_OPENAI_API_KEY in the main function with your actual OpenAI API key.

How to Run

1. Ensure you have all the required datasets in the specified paths.
2. Run the script by executing the following command:

python movie_recommendation_system.py

Function Descriptions

load_datasets

Loads the necessary datasets for the project.

preprocess_movies

Preprocesses the movies dataset by removing movies with no genres listed and merging with the links dataset.

cluster_movies

Performs TF-IDF vectorization on movie titles and genres and applies K-means clustering.

select_movies_from_clusters

Selects one movie from each cluster.

get_recommendations

Uses the OpenAI API to generate movie recommendations based on the selected movies.

check_present

Checks if the predicted movie is present in the dataset.

map_title_to_id

Maps predicted movie titles to their respective movie IDs.

main

Main function to execute the workflow of loading data, preprocessing, clustering, selecting movies, getting recommendations, and calculating percentages.

Output

The final output is saved to output1.csv which contains the recommended movies and their respective ratings.

Notes

• The script calculates the percentage of users whose ratings for the recommended movies are within a threshold of 0.5 from their original ratings.
• Ensure the OpenAI API key is kept secure and not shared publicly.