Test for Collaborative Filtering

For an example, let's suppose the following data:

• There are 6 products (R1,R2,R3,R4,R5 and R6);
• There are 3 "training set" users (Bob, Chris and Diana);
• The query instance is Alice's ratings on the 5.

User/Product	R1	R2	R3	R4	R5	R6
Alice	2	-	4	4	-	5
Bob	1	5	4	-	3	4
Chris	5	2	-	2	1	-
Diana	3	-	2	2	-	4

But the data does not come in this table format. Actually, let's imagine a log file where each row are: movieID, customerID,rating. For simplicity, let's suppose Alice's ID 1, Bob's ID 2, Chris' ID 3 and Diana's 4.

Our training data, then, come in the following format. Notice since Alice is the query instance, she does not appear in the training data.

movieID	customerID	rating
1	2	1
2	2	5
3	2	4
5	2	3
6	2	4
1	3	5
2	3	2
4	3	2
5	3	1
1	4	3
3	4	2
4	4	2
6	4	4

Query instance (Alice's logs) is:

movieID	customerID	rating
1	1	2
3	1	4
4	1	4
6	1	5

What we want to know (to predict) is Alice's rate on movie 5. We can easily estimate this value using mtxslv_collab_filter(). Let's define training and testing data:

training_set_example = np.array([[1,2,1],[2,2,5],[3,2,4],[5,2,3],
                         [6,2,4],[1,3,5],[2,3,2],[4,3,2],
                         [5,3,1],[1,4,3],[3,4,2],[4,4,2],
                         [6,4,4]])

testing_instance = np.array([[1,1,2],[3,1,4],[4,1,4],[6,1,5]])

Now apply the function!

estimativa = mtxslv_collab_filter(testing_instance,5,training_set_example)

The manually calculated value is 4.336075363. Test it yourself, estimativa is 4.336096188777122.