Movie Recommendation System using NLTK library

In this project the TMDB data set (https://www.kaggle.com/datasets/tmdb/tmdb-movie-metadata) has been used to build a movie recommendation system.

'CountVectorizer' has been used to perform the Text to Vector transformation.

This project is deployed using 'Streamlit'.

How to run ?

• Download the whole package as zip.
• Unzip the package, i.e. 'movie_recommendation.zip':

unzip movie_recommendation.zip

• Go to the new folder (ie. the folder created after unzipping)

cd movie_recommendation

• Create a local enviornment. I have used python version 3.8. One can use higher version too.

conda create -p venv python==3.8 -y

• Activate the new enviornment.

conda activate venv/

• Now make a directory named 'artifacts'.

mkdir artifacts

• To install required libraries, do -

pip install -r requirements.txt

• Now you can run the notebook file - 'notebooks/movie_reco.ipynb'.
• This will generate the 'movie_data.pkl' which is the input list of names of movies and 'cosine_similarity.pkl' which stores the cosine similarity matrix.
• These above two files will be needed when we run 'app.py' using 'Streamlit'.
• To run 'app.py' -

streamlit run app.py

This will open a browser at localhost where you will be able to get the recommendations.

Here is a glimpse of the output.

• Run it and feel free to ask any doubts. Happy coding !!