FScore with best threshold calculation

[PedroCardoso]

Describe the feature and the current behavior/state.
A F1Score metric that allowed us to define N threshold splits. We would calculate values to each threshold, and onn result we would return the best F Score.
This could be an update to the existing FBetaScore or a new class. On the existing class, we would have a num_thresholds argument to init . result() would work as now, returning the best, and we could add a new method best_threshold()

Relevant information

• Are you willing to contribute it (yes/no): yes
• Are you willing to maintain it going forward? (yes/no): yes
• Is there a relevant academic paper? (if so, where): no
• Is there already an implementation in another framework? (if so, where): It used to exist in TF1
• Was it part of tf.contrib? (if so, where): There was metrics on TF1 called *at_thresholds that worked similarly , I believe.

Which API type would this fall under (layer, metric, optimizer, etc.)
metric

Who will benefit with this feature?
me ! and anyone wanting best F Score at best threshold !

Any other info.

[bhack]

/cc @SSaishruthi @marload

[marload]

This feature seems useful to me. Personally, it seems better to create a new class than to update FBetaScore. 😄

[PedroCardoso]

I can take a go, but would be happy to have proposals for Class names

[seanpmorgan]

TensorFlow Addons is transitioning to a minimal maintenance and release mode. New features will not be added to this repository. For more information, please see our public messaging on this decision:
TensorFlow Addons Wind Down

Please consider sending feature requests / contributions to other repositories in the TF community with a similar charters to TFA:
Keras
Keras-CV
Keras-NLP