Movie Recommender recommends you top five movies based on a single entry of a movie. The project explores content based filtering of data, which is a type of recommender system that attempts to guess what a user may like based on that userβs activity. Content-based filtering makes recommendations by using keywords and attributes assigned to objects in a database. Youtube uses CBF along with other filtering techniques. The backend service is provided by streamlit and is hosted on Heroku.
The different stages of the project included:
- Preprocessing where the data from two csv files which contained information about 4086 movies were processed for futher detailing.
- Model building where a model was trained using the algorithm of k-nearest neighbours using CountVectorizer and cosine_similarity libraries.
- Website building done using streamlit
- Deployed on Heroku
Click here to view the deployment
- Python 2.7 or greater
- Natural Language Toolkit (NLTK)
- Streamlit
- Pickle
- Requests
- Sci-Kit learn
- Numpy
and a few other libraries and dependencies for preprocessing.
- Opening Screen
- Option Menu - I
- Recommendations - I
- Recommendations - II
π Click Here to test the application on your own.
π If you liked the application, make sure to star this repo, Thankyou.