How does surprise library handle missing values in matrix factorisation SVD model?
I am not sure how it handles missing values. Does it fill NA values with 0 and then do the matrix factorization or it simply does not consider those missing values while calculating cost function.