search

fidelity / mab2rec

[AAAI 2024] Mab2Rec: Multi-Armed Bandits Recommender
https://fidelity.github.io/mab2rec/
129 stars 26 forks source link

LinUCB returning NaN in predicted rewards/ response #17

Closed ayush488 closed 2 years ago

ayush488 commented 2 years ago

What could be the reason that LinUCB is returning NaN in the predicted rewards? LinGreedy is working fine with the same data.

bkleyn commented 2 years ago

If you provide a minimal example (code and data) to reproduce this error we can have a closer look.

My only guess here would be to make sure that you have enough observations for each arm/item. For LinUCB variance is undefined if you were to have only one observation for a given arm/item for example.

ayush488 commented 2 years ago

Got..it.. so that would be the problem.. My data does not have " enough observations for each arm/item"

Regards, Ayush

On Mon, Jul 18, 2022 at 7:49 AM Bernard Kleynhans @.***> wrote:

If you provide a minimal example (code and data) to reproduce this error we can have a closer look.

My only guess here would be to make sure that you have enough observations for each arm/item. For LinUCB variance is undefined if you were to have only one observation for a given arm/item for example.

— Reply to this email directly, view it on GitHub https://github.com/fidelity/mab2rec/issues/17#issuecomment-1187332085, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDPIEU4CAJO32J36AEEOQTVUVHEVANCNFSM53WS74XA . You are receiving this because you authored the thread.Message ID: @.***>