Hello to all!
I have a question about the baseline predictors method.
I have a sum of ratings for each user and a sum of ratings for each item. Now I try to predict how a certain user will rate an item. I do it like this
raw_average + bias (пользователя) + bias (элемента)
Every bias i find like
(сумма всех оценок конкретным пользователем (или сумма всех оценок конкретного элемента) / кол-во оценок) - raw_average
But there are users who can not rate a single item. The same goes for the element (Someone else can not evaluate any element).
And then what? Then the кол-во элементов == 0 in bias ..., And I need to divide by zero ...
