Skip to content

/ scikit-learn Public

New issue

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

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

Already on GitHub? Sign in to your account

Jump to bottom

EHN Provide a pos_label parameter in plot_precision_recall_curve #17569

Merged
merged 12 commits into from Jun 24, 2020
Merged

EHN Provide a pos_label parameter in plot_precision_recall_curve #17569

merged 12 commits into from Jun 24, 2020

Conversation

@glemaitre
Copy link
Contributor

@glemaitre glemaitre commented Jun 11, 2020 •
edited

closes #17565
partially addressing #15573

Add a pos_label parameter to specify which class to be the positive class when estimator.classes_[1] is not the right choice.
@glemaitre
Provide a pos_label parameter in plot_precision_recall_curve
c5d0989
@glemaitre glemaitre changed the title Provide a pos_label parameter in plot_precision_recall_curve EHN Provide a pos_label parameter in plot_precision_recall_curve Jun 11, 2020
@glemaitre
PEP8
01287f8
@thomasjpfan
thomasjpfan reviewed Jun 11, 2020
View changes
Copy link
Member

@thomasjpfan thomasjpfan left a comment •
edited

Thank you @glemaitre ! LGTM

Edit: Dose not work with response_method='decision_function'
@thomasjpfan thomasjpfan dismissed their stale review Jun 11, 2020

Testin Failing T_T

@ogrisel
Copy link
Member

@ogrisel ogrisel commented Jun 11, 2020

Should we display the pos label in the axis labels or in the title of the plot?

@thomasjpfan
Copy link
Member

@thomasjpfan thomasjpfan commented Jun 11, 2020

Do we restrict this to response_method='predict_proba'?

@glemaitre
Copy link
Contributor Author

@glemaitre glemaitre commented Jun 11, 2020

Should we display the pos label in the axis labels or in the title of the plot?

I think it would be great indeed.

Do we restrict this to response_method='predict_proba'?

I did not think yet at the decision function but I think that we should come with a fix as well. I am looking at it now.

@glemaitre
Copy link
Contributor Author

@glemaitre glemaitre commented Jun 11, 2020

Basically we will be fine with decision_function since we are giving the y_pred which is 1D and pos_label to the precision_recall_curve function which will manage properly the issue with those.

The probability is more problematic because precision_recall_curve should be a 1D array and we have to do the job before to pass the function.

@glemaitre
fix pos_label to work with both predict_proba and decision_function
9be50f0
@glemaitre
Copy link
Contributor Author

@glemaitre glemaitre commented Jun 11, 2020

I think that the question of @amueller in #15573 should be addressed. Is pos_label the right parameter name with its slightly semantic being different from the precision_recall_curve for instance.

@glemaitre
add positive label in axis
a39d41a
@thomasjpfan
Copy link
Member

@thomasjpfan thomasjpfan commented Jun 11, 2020

We are using pos_label to do two things here:

  1. Select the positive class from classes_
  2. Passing to precision_recall_curve which selects the positive class from y_true.

So in the sense of 2, we have the same semantics. I think the semantics for 1 is clear enough compared to the cost of introducing another parameter name.

@glemaitre
Copy link
Contributor Author

@glemaitre glemaitre commented Jun 11, 2020

Fair enough with me. Would we consider this as a bug fix or an enhancement?

@thomasjpfan
Copy link
Member

@thomasjpfan thomasjpfan commented Jun 11, 2020

Adding a new parameter feels like an enhancement.

@glemaitre
add entry in what new
c7f6824
@thomasjpfan
thomasjpfan reviewed Jun 11, 2020
View changes
sklearn/metrics/_plot/precision_recall_curve.py Show resolved Hide resolved
glemaitre added 2 commits Jun 14, 2020
@glemaitre
iter
bc347ab
@glemaitre
PEP8
be9e86d
@glemaitre glemaitre mentioned this pull request Jun 15, 2020
Closed
@glemaitre
add versionadded
ffab343
@glemaitre glemaitre added this to WAITING FOR REVIEW in Guillaume's pet Jun 17, 2020
@thomasjpfan
thomasjpfan approved these changes Jun 17, 2020
View changes
Copy link
Member

@thomasjpfan thomasjpfan left a comment

LGTM
sklearn/metrics/_plot/precision_recall_curve.py Show resolved Hide resolved
glemaitre added 2 commits Jun 17, 2020
@glemaitre
Update precision_recall_curve.py
f7bf1fe
@glemaitre
PEP8
9edc072
@thomasjpfan
thomasjpfan reviewed Jun 17, 2020
View changes
sklearn/metrics/_plot/tests/test_plot_precision_recall.py Show resolved Hide resolved
@claramatos
Copy link
Contributor

@claramatos claramatos commented Jun 20, 2020

is this ready to be closed? if so can someone please close it? (@thomasjpfan) I'm working on #15573 and would like to make my changes on top of these ones

@thomasjpfan
Copy link
Member

@thomasjpfan thomasjpfan commented Jun 20, 2020

This needs a second reviewer to approve before it can get merged.

@claramatos
Copy link
Contributor

@claramatos claramatos commented Jun 20, 2020

so what is the best thing to do? should I merge this branch into mine? or should I make my changes and then solve possible conflicts once this branch is approved?

@amueller
Copy link
Member

@amueller amueller commented Jun 20, 2020

@claramatos the second indeed. Just make your changes against master for now and later resolve conflicts. You could also base your PR on this one but github doesn't have good PR dependency tracking, so this is a bit hard to disentangle.

@claramatos claramatos mentioned this pull request Jun 21, 2020
Merged
@glemaitre
Copy link
Contributor Author

@glemaitre glemaitre commented Jun 22, 2020

I'll be a bit annoying here. Anybody able to make a second review and a potential merge @ogrisel @jeremiedbb @amueller @NicolasHug

@ogrisel
ogrisel approved these changes Jun 24, 2020
View changes
Copy link
Member

@ogrisel ogrisel left a comment

LGTM as well. Just a nit. I will merge when green.
sklearn/metrics/_plot/tests/test_plot_precision_recall.py Outdated Show resolved Hide resolved
@ogrisel
Update sklearn/metrics/_plot/tests/test_plot_precision_recall.py
6e6d2ff
@ogrisel ogrisel mentioned this pull request Jun 24, 2020
Merged
@ogrisel
Merge branch 'master' into is/17565
93e3530
@ogrisel ogrisel merged commit e5b99ea into scikit-learn:master Jun 24, 2020
3 of 4 checks passed
3 of 4 checks passed
ci/circleci: lint CircleCI is running your tests
Details
@lgtm-com
LGTM analysis: C/C++ No code changes detected
Details
@lgtm-com
LGTM analysis: JavaScript No code changes detected
Details
@lgtm-com
LGTM analysis: Python No new or fixed alerts
Details
rubywerman added a commit to MLH-Fellowship/scikit-learn that referenced this pull request Jun 24, 2020
@glemaitre @rubywerman
EHN Provide a pos_label parameter in plot_precision_recall_curve (sci…
b40b1db 
…kit-learn#17569)
@glemaitre glemaitre moved this from WAITING FOR REVIEW to WAITING FOR CONSENSUS in Guillaume's pet Jun 25, 2020
@glemaitre glemaitre moved this from WAITING FOR CONSENSUS to MERGED in Guillaume's pet Jun 25, 2020
jayzed82 added a commit to jayzed82/scikit-learn that referenced this pull request Oct 22, 2020
@glemaitre @jayzed82
EHN Provide a pos_label parameter in plot_precision_recall_curve (sci…
52422ca 
…kit-learn#17569)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@thomasjpfan thomasjpfan

@ogrisel ogrisel

Assignees
No one assigned
Labels
module:metrics
Projects
Guillaume's pet
MERGED
Milestone
No milestone
Linked issues

Successfully merging this pull request may close these issues.

plot_precision_recall_curve should expose pos_label
5 participants
@glemaitre @ogrisel @thomasjpfan @claramatos @amueller