Recommender System in Python - II (optional) with Ecommerce Data Science

Now, let's learn about item-based recommendation.

Item-based similarity is similar to user-based similarity. Here, you need to find the correlation between different items and, based on that, you recommend items to the user. Here, the item correlation is of shape 2500 x 2500. Multiplying by the transpose of the 'movie_feature' (rating by the user) of shape 862 x 2500 and, again, taking the transpose will give the item-based recommendation of the users, which is of shape 862 x 2500. Now, let's evaluate the recommendation system that you have developed.

As you saw, we follow the same procedure to evaluate the user-based as well as the item-based recommendation system. Here, 'dummy_ test' is used to keep only the movie ratings given by the user. RMSE is used as evaluation metrics.