bugbug icon indicating copy to clipboard operation
bugbug copied to clipboard

more automatic labelled examples for the stepstoreproduce model

Open chidauri opened this issue 6 years ago • 5 comments

Fixes #426

chidauri avatar Jul 08 '19 13:07 chidauri

Codecov Report

Merging #681 into master will increase coverage by 45.21%. The diff coverage is n/a.

Impacted file tree graph

@@            Coverage Diff            @@
##           master   #681       +/-   ##
=========================================
+ Coverage   54.78%   100%   +45.21%     
=========================================
  Files          77      1       -76     
  Lines        5251     14     -5237     
=========================================
- Hits         2877     14     -2863     
+ Misses       2374      0     -2374
Impacted Files Coverage Δ
tests/test_assignee.py
bugbug/models/component.py
bugbug/bug_features.py
scripts/commit_retriever.py
bugbug/bugzilla.py
bugbug/bug_snapshot.py
tests/test_devdocneeded.py
bugbug/__init__.py
bugbug/models/defect_enhancement_task.py
scripts/regressor_finder.py
... and 64 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 0e28080...f77774e. Read the comment docs.

codecov-io avatar Jul 08 '19 13:07 codecov-io

I think you might be rollbacking all bugs this way, are you sure you aren't?

marco-c avatar Jul 08 '19 15:07 marco-c

I think you might be rollbacking all bugs this way, are you sure you aren't?

https://github.com/mozilla/bugbug/blob/b456bd3b2d00aad453b42cc8761fc04fbf0e3866/bugbug/bug_features.py#L444-L449 This helped :) PS. This is why we yield the bugs (which we want the rollbacked version of), first.

chidauri avatar Jul 08 '19 16:07 chidauri

I think relying on that is a little brittle:

  1. there might be cases when you want to rollback the same bug multiple times in different ways;
  2. you are going to rollback the bug the first time it appears, so for example for the stepswanted case you might rollback the bug with label 1 while you should have rollbacked the bug with label 0. Right now it doesn't happen as you are yielding the bugs to rollback first, but it's brittle to rely on this.

marco-c avatar Jul 08 '19 16:07 marco-c

Before changes

73 bugs have no steps to reproduce 2642 bugs have steps to reproduce X: (2715, 68672), y: (2715,) Cross Validation scores: Accuracy: f0.6930046232725715 (+/- 0.08506114027939689) Precision: f0.9889862871434468 (+/- 0.011123332687863755) Recall: f0.6927315347191507 (+/- 0.09585654167878219) X_train: (128, 68672), y_train: (128,) X_test: (272, 68672), y_test: (272,)

After changes

73 bugs have no steps to reproduce 2642 bugs have steps to reproduce X: (2730, 68473), y: (2730,) Cross Validation scores: Accuracy: f0.6882049784793953 (+/- 0.061864732077525464) Precision: f0.9866638261388336 (+/- 0.009726157216997822) Recall: f0.6885503013078603 (+/- 0.05885443628645505) X_train: (136, 68473), y_train: (136,) X_test: (273, 68473), y_test: (273,)

chidauri avatar Jul 14 '19 19:07 chidauri