Add Specificity metric

[arv-77]

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 #106 .
Added classification and functional for specificity. Also added tests

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 🙃

[pep8speaks]

Hello @arvindmuralie77! Thanks for updating this PR.

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2021-05-03 17:25:02 UTC

[maximsch2]

Can we make the testing simpler by flipping the labels and predictions and calculating regular recall (by calling sklearn functions)

[SkafteNicki]

Overall LGTM, some extra comments:

• remember to add entry to the docs/source/references/modules.rst and docs/source/references/functional.rst
• remember to add entry to CHANGELOG.md file

[codecov]

Codecov Report

Merging #210 (a467b73) into master (e46a7d2) will decrease coverage by 0.05%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #210      +/-   ##
==========================================
- Coverage   96.68%   96.62%   -0.06%     
==========================================
  Files         180       92      -88     
  Lines        5786     2935    -2851     
==========================================
- Hits         5594     2836    -2758     
+ Misses        192       99      -93     

Flag	Coverage Δ
Linux	79.76% <85.36%> (+0.07%)	⬆️
Windows	79.76% <85.36%> (+0.07%)	⬆️
cpu	96.62% <100.00%> (-0.06%)	⬇️
gpu	?
macOS	96.62% <100.00%> (-0.06%)	⬇️
pytest	96.62% <100.00%> (-0.06%)	⬇️
python3.6	?
python3.8	96.59% <100.00%> (-0.06%)	⬇️
python3.9	96.59% <100.00%> (+0.04%)	⬆️
torch1.3.1	?
torch1.4.0	?
torch1.8.1	96.59% <100.00%> (+0.04%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
torchmetrics/__init__.py	100.00% <ø> (ø)
torchmetrics/classification/__init__.py	100.00% <100.00%> (ø)
torchmetrics/classification/specificity.py	100.00% <100.00%> (ø)
torchmetrics/functional/__init__.py	100.00% <100.00%> (ø)
torchmetrics/functional/classification/__init__.py	100.00% <100.00%> (ø)
...chmetrics/functional/classification/specificity.py	100.00% <100.00%> (ø)
torchmetrics/average.py	83.33% <0.00%> (-12.50%)	⬇️
__w/2/s/torchmetrics/utilities/__init__.py
...2/s/torchmetrics/retrieval/mean_reciprocal_rank.py
...unctional/classification/precision_recall_curve.py
... and 89 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e46a7d2...a467b73. Read the comment docs.

[SkafteNicki]

Hi @arvindmuralie77,
Code is looking great. Can I request one more change? We are in the process of adding test for differentiability, so it would be great if you could:

• Add this property to the metric: https://github.com/PyTorchLightning/metrics/blob/263a4c7a651f4c85cf79003d59328c08430020bd/torchmetrics/classification/accuracy.py#L275-L277
• Add the corresponding test: https://github.com/PyTorchLightning/metrics/blob/263a4c7a651f4c85cf79003d59328c08430020bd/tests/classification/test_accuracy.py#L106-L116

[SkafteNicki]

LGTM, only change is to remove the unused arguments. I know that other metrics are doing this, but lets try not to do it.

[Borda]

lgtm