Collaborative Filtering Prediction

Problem

Predict a missing rating with user-based collaborative filtering.

For the target user, average the ratings other users gave the item, weighted by their cosine similarity to the target: sum(sim * rating) / sum(abs(sim)) over users with a nonzero rating, excluding the target user. Return 0.0 when no other user qualifies.

Hint

Only neighbors who actually rated the item contribute to the weighted average.

Test cases

case 1

in: [[[5,0],[4,4]],[[1,0.5],[0.5,1]],0,1]

expected: 4

case 2

in: [[[3,0],[0,5]],[[1,0],[0,1]],0,1]

expected: 0

case 3

in: [[[0,0],[2,0],[0,4]],[[1,0.8,0.2],[0.8,1,0.5],[0.2,0.5,1]],0,0]

expected: 2

case 4

in: [[[4,0],[0,0]],[[1,0],[0,1]],0,1]

expected: 0

Starter codePython

def cf_predict(matrix, similarity, user, item):
    # Your code here
    pass

4 test cases · run your code against them in the editor