Adds average argument to AveragePrecision metric

[SkafteNicki]

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?

What does this PR do?

Fixes #471
Adds average argument to AveragePrecision metric as requested by users. This will change the default output for multiclass and multilabel input from a list with the score per class to instead output the macro average.

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[Borda]

@SkafteNicki seems average_precision now returns single number instead of tuple

[codecov]

Codecov Report

Merging #477 (6e41b91) into master (043714e) will decrease coverage by 0%.
The diff coverage is 94%.

@@          Coverage Diff          @@
##           master   #477   +/-   ##
=====================================
- Coverage      95%    95%   -0%     
=====================================
  Files         132    132           
  Lines        4652   4681   +29     
=====================================
+ Hits         4435   4459   +24     
- Misses        217    222    +5     

[SeanNaren]

nice!

[SeanNaren]

fine with the wording choice here, but isn't mean usually the term? i.e mAP?

[Borda]

fine with the wording choice here, but isn't mean usually the term? i.e mAP?

good point, mAP is mean Average Precision but it is not simple stat mean, but some aggregated mean if I am correct 🐰
@SeanNaren pls have also look at #467

[SkafteNicki]

This is the confusing part with the naming of these metrics:

• AveragePrecision (the metric in this PR) refers to the area under the precision-recall curve evaluated user different thresholds. So the name basically means: the average value of precision given all the full range of recall values
• MeanAveragePrecsion or mAP most commonly refers to metric used in object detection. It is also related to the average precision, but the mean is calculated instead over different thresholds of intersection over union between objects (as far as I understand)
  So "same same but different".

[SeanNaren]

Makes sense, thanks for the clarification guys :)