[REFACTOR] python evaluation no longer use threshold

[loomlike]

Description

• Python evaluation seems no longer use threshold-based relevancy calculation. Should remove the unused argument from functions.
• Spark evaluation still have both timestamp and threshold-based relevancy calculation.

In which platform does it happen?

• python
• pyspark

Other Comments

Few more clean-up:

• merge_ranking_true_pred function no longer use col_rating
• the function is also missing argument descriptions of the last three parameters

[yueguoguo]

Python evaluation seems no longer use threshold-based relevancy calculation. Should remove the unused argument from functions.

Maybe it is rationale to still keep this as the evaluation parts in all of the notebooks in the repo follow the top k recommendation protocol. This does not mean there is no use case for other types of recommendation scenarios, i.e., recommending items that have predicted ratings larger than a pre-defined threshold.

Spark evaluation still have both timestamp and threshold-based relevancy calculation.

I can add the Python version.

Thoughts? @loomlike

merge_ranking_true_pred function no longer use col_rating

It is used by the relevancy function (if we keep it)

the function is also missing argument descriptions of the last three parameters

Will add them