This project is based on Content-Based Filtering, which utilizes item metadata such as genre, director, description, actors, etc., for movies to provide recommendations. The underlying principle of these recommender systems is that if a person enjoys a particular item, they are likely to appreciate items similar to it.

1. Data Collection: Data was sourced from Kaggle. Following a meticulous selection process, only specific columns were retained while the rest were discarded:

   • id
   • title
   • genres
   • keywords
   • overview
   • cast
   • crew

2. Data Preprocessing: Textual data underwent preprocessing where they were concatenated into a tags column, serving as the foundation for subsequent modeling.

3. Text Processing: Tags were subjected to stemming, tokenization, and vectorization (Bag of Words). Distances were computed using Cosine similarity.

4. Recommender System: A recommender function was developed to identify movies similar to a given film.

5. Application Development: The processed files were pickled and utilized in app.py to create an application using Streamlit.

6. Deployment: Initially, the application was hosted on Heroku.

By following this workflow, the Movie Recommender System was developed, allowing users to discover movies based on their preferences and similarities.