Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

[WIP ]Added labelwise metrics with tests #542

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

anmolsjoshi
Copy link
Contributor

Fixes #513 and #516

Description:
Added tests

Check list:

  • New tests are added (if a new feature is added)
  • New doc strings: description and/or example code are in RST format
  • Documentation is updated (if required)

@anmolsjoshi anmolsjoshi requested a review from vfdev-5 June 5, 2019 04:04
@anmolsjoshi
Copy link
Contributor Author

@vfdev-5 I have at least added Accuracy with labelwise option. Wanted to further discuss Precision and Recall before implementing.

Copy link
Collaborator

@vfdev-5 vfdev-5 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not a fan of another bool flag labelwise. Idea is to introduce a new variable, e.g multilabel that takes values for example exact, labelwise, top-K with variable K (top-3, top-5 etc):

  • exact is current is_multilabel=True behaviour
  • labelwise is your implementation behaviour with labelwise=True
  • top-K with parsing K into an integer and computes top-k multi-label accuracy.

@anmolsjoshi what do you think ?

Old argument is_multilabel becomes deprecated and will be removed since 0.3.0

def __init__(self, output_transform=lambda x: x, is_multilabel=False):
def __init__(self, output_transform=lambda x: x, is_multilabel=False, labelwise=False):
if labelwise and not is_multilabel:
raise ValueError("labelwise is only applicable for multilabel cases. If labelise=True, is_multilabel=True.")
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
raise ValueError("labelwise is only applicable for multilabel cases. If labelise=True, is_multilabel=True.")
raise ValueError("labelwise is only applicable for multilabel cases. If labelwise=True, is_multilabel=True.")

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Label-wise metrics (Accuracy etc.) for multi-label problems
2 participants