Finding similarly liked movies
Just like you calculated the similarity between two movies, you can calculate it across all users to find the most similar movie to another based on how users have rated them.
The approach is similar to how you worked with content-based filtering.
You will find the similarity scores between all movies and then drill down on the movie of interest by isolating and sorting the column containing its similarity scores.
movie_ratings_centered has once again been loaded, containing each movie as a row, and their centered ratings stored as the values.
This exercise is part of the course
Building Recommendation Engines in Python
Exercise instructions
- Calculate the similarity matrix between all movies in
movie_ratings_centeredand store it as similarities. - Wrap the
similaritiesmatrix in a DataFrame, with the indices ofmovie_ratings_centeredas the columns and rows.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
from sklearn.metrics.pairwise import cosine_similarity
# Generate the similarity matrix
similarities = ____(____)
# Wrap the similarities in a DataFrame
cosine_similarity_df = ____.____(____, index=____.____, columns=____.____)
# Find the similarity values for a specific movie
cosine_similarity_series = cosine_similarity_df.loc['Star Wars: Episode IV - A New Hope (1977)']
# Sort these values highest to lowest
ordered_similarities = cosine_similarity_series.sort_values(ascending=False)
print(ordered_similarities)