16. Video: Identifying Recommendations
Identifying Recommendations
Making Recommendations
Finalizing our neighborhood based recommendations, we need to use the ratings from our neighbors to influence the ratings we provide to other users.
There are a few ways to do this, but a simple method would be to:
- Remove movies our user has already seen.
- Find ratings of the neighbors that are high.
- Recommend movies to each user where both 1 and 2 above hold.
Other methods for making recommendations using collaborative filtering are based on weighting of the neighbors' ratings based on the 'closeness' of the neighbors.
You can use each of the following two papers to learn more about this technique:
In the next notebook, you will implement the three-step process above to make recommendations for every user in the dataset. For computational reasons, you will notice that iterating this approach through all users has been done for you. But you will go through the process of implementing for individual pairs of users, which could easily be extended via looping to all users.