[FEATURE] Efficiency enhancement in surprise_utils.py

[zhaisw]

Description

In recommenders/reco_utils/recommender/surprise/surprise_utils.py compute_ranking_predictions function, the code will try to get unique items or users in every for-loop step, which will decrease the efficiency.

Expected behavior with the suggested feature

Better to use memory to improve code efficiency,

users = data[usercol].unique()
items = data[itemcol].unique()
for user in users:
        for item in items:
...

Other Comments

[miguelgfierro]

this is really nice, do you want to do a PR?

[zhaisw]

this is really nice, do you want to do a PR?

Yes! For sure, I would like to do that. Thanks.

[miguelgfierro]

thanks for the contribution @zhaisw!
#1224