movie-recommender

Movie recommender web application for the MovieLens 1M dataset. The repository was built along with @sarabase, @sergibech, @Goodjorx & @MartaBuetas as implementation of the AGILE DS's course project of the Data Science MSc (UB, 2022-23) course.

Repository folder structure

• The .streamlit folder: contains the Streamlit configuration file including the theme.
• The .github folder: contains instructions regarding the CI/CD process.
• The src folder: contains the main code in python (the soul of our Streamlit app, what it displays, how it reacts to user input...)
• The assets folder: contains images necessary for the app frontend.
• The data folder: contains the model serialization
• The docs folder: contains the necessary files to generate the documentation

Commands

Important: in order to be able to properly run the web app, one must ensure it exists the .credentials hidden file in the backend/app subdirectory. If you don't have the file, please refer to @gcastro-98 as database's administrator.

Locally run everything

We can locally test our web application by creating docker image and launching the corresponding docker container through the following command:

docker-compose up --force-recreate --no-deps --build

Alternatively, we can run it without docker if we have installed the requirements in a conda environment using:

pip install -r requirements.txt

And then, executing the web app using:

make run

Generate package documentation

First we need to have a conda environment in which install all the necessary packages to 'compile' the src code, by typing:

pip install -r requirements.txt

Afterwards, the necessary packages to generate the documentation must be installed as:

conda install -c anaconda sphinx numpydoc sphinx_rtd_theme recommonmark -y
conda install -c anaconda python-graphviz openpyxl -y
pip install --upgrade myst-parser

Also, in the docs folder, there must exist a copy of the data folder, as well as the same .credentials file in the docs/src subdirectory. Finally, we simply type:

sphinx-build docs/src docs/build

or equivalently:

make html

Then, opening the vortexpy/docs/build/index.html file in the browser will display the generated documentation.