Metadata-based recommender
We will largely follow the same steps as the plot description-based recommender to build our metadata-based model. The main difference, of course, is in the type of data we use to build the model.
Preparing the data
To build this model, we will be using the following metdata:
- The genre of the movie.
- The director of the movie. This person is part of the crew.
- The movie's three major stars. They are part of the cast.
- Sub-genres or keywords.
With the exception of genres, our DataFrames (both original and cleaned) do not contain the data that we require. Therefore, for this exercise, we will need to download two additional files: credits.csv, which contains information on the cast and crew of the movies, and keywords.csv, which contains information on the sub-genres.
Note
You can download the necessary files from the following URL: https://www.kaggle.com/rounakbanik/the-movies-dataset/data.
Place both files in your data folder. We need to perform a good amount of wrangling...
