Question about Top N > 100 about qrec

[Coder-Yu]

The length of itemSet[item] is less than N because some items which appear in the training set have been removed from the itemSet. So it is natural to get an IndexError.
You could change 100 to a larger value, however, it is wrong to set it to the number of items. Changing base class is not suggested unless you can improve the effectiveness or need new features.

[statusrank]

The length of itemSet[item] is less than N because some items which appear in the training set have been removed from the itemSet. So it is natural to get an IndexError.
You could change 100 to a larger value, however, it is wrong to set it to the number of items. Changing base class is not suggested unless you can improve the effectiveness or need new features.

Thansks!
"Items which appear in training dataset are removed from the itemSet".This means ,if the user_u have $item_i$ in the training set，when we run our model on test set， we should ignore $item_i$ for user_u?
In my experiment,for per user,I need sort all items except items which is appeared in the training set for this user.
If the RecQ can do this， how do i know the number of items left for per user? thanks

[Coder-Yu]

Actually, no one cares the number of items left for per user. In your experiments, you just need to calculate the average performance. Paying much attention to single user is meaningless.