Always use IDs in plots and brain results #2614
Conversation
Codecov ReportBase: 62.53% // Head: 62.46% // Decreases project coverage by
Additional details and impacted files@@ Coverage Diff @@
## develop #2614 +/- ##
===========================================
- Coverage 62.53% 62.46% -0.08%
===========================================
Files 249 249
Lines 42168 42262 +94
Branches 347 347
===========================================
+ Hits 26371 26399 +28
- Misses 15797 15863 +66
Flags with carried forward coverage won't be shown. Click here to find out more.
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
| patches_field, leaf | ||
| ) | ||
|
|
||
| labels = curr_view.values(label_field, unwind=True) | ||
| field = curr_view.get_field(label_field) | ||
| labels = view._get_values_by_id( |
There was a problem hiding this comment.
Updates embeddings backend to always pull color-by data using IDs
| ) | ||
|
|
||
| if ids is not None and not is_frames: | ||
| values = samples._get_values_by_id( |
There was a problem hiding this comment.
All of the refactoring in the plotting methods was to achieve this line: when the user has provided IDs and is asking to pull values from a dataset via path/expression, use _get_values_by_id() to ensure that the correct values are pulled, even if samples doesn't correspond 1-1 with other data that may have been provided.
| plot = fo.scatterplot( | ||
| points=points, | ||
| samples=dataset, | ||
| ids=ids, |
There was a problem hiding this comment.
This is an example of something that wouldn't previously work. User provided ids + points corresponding to a subset of the samples=dataset argument that they provided, but they provided paths/expressions for labels and sizes arguments.
Previously this would fail because dataset.values() would naively be used (which would result in too many labels/sizes). Now the ids argument is used to lookup the correct labels/sizes in the correct order corresponding to points/ids.
Updates methods like
scatterplot()and the in-App embeddings backend to always rely on sample/label IDs when pulling data for plots.Previously, these implementations assumed that the
viewon which embeddings/etc were computed has not changed in any way. Now, these methods will gracefully handle added/deleted data.New tests are added to
tests/intensive/plot_tests.pyto verify that the plotting methods function as expected.