How to generate negative examples collaberative filtering'?
Answers
Answered by
0
if the predicted value is below the minimum value of the rating scale, set it to the minimum value (1 in this case), and for the predicted values > max of the rating scale, set them to the max rating (5 in this case).
When you're evaluating a CF algorithm, this hack would mess up a bit of course, but this works well for a live system. Also, the hack makes sense in terms of the goals of the predictions.
When you're evaluating a CF algorithm, this hack would mess up a bit of course, but this works well for a live system. Also, the hack makes sense in terms of the goals of the predictions.
Similar questions