API Adds defaults to Display Objects by thomasjpfan · Pull Request #16933 · scikit-learn/scikit-learn

thomasjpfan · 2020-04-15T17:52:32Z

Reference Issues/PRs

Partially addresses #15880 with more docs.

What does this implement/fix? Explain your changes.

This PR adds

Defaults to the display object to make it easier to construct.
Example that shows how to create the display objects and plot them.

…ample

rth

This is nice, thanks @thomasjpfan !

adrinjalali

Nice one.

examples/plot_display_object_visualization.py

NicolasHug

This is nice!

Nits but LGTM when addressed

NicolasHug · 2020-04-21T18:21:32Z

sklearn/metrics/_plot/confusion_matrix.py

+        Display labels for plot. If None, display labels are set to 0 to
+        `n_classes`.


Suggested change

Display labels for plot. If None, display labels are set to 0 to

`n_classes`.

Display labels for plot. If None, display labels are set from 0 to

`n_classes - 1`.

NicolasHug · 2020-04-21T18:25:45Z

sklearn/metrics/_plot/confusion_matrix.py

+        if self.display_labels is None:
+            display_labels = np.arange(n_classes)
+        else:
+            display_labels = self.display_labels


Do we like that kind of idiom? Just wondering

display_labels = self.display_labels or np.arange(n_classes)

We can not do this because self.display_labels can be a numpy array.

NicolasHug · 2020-04-21T18:28:52Z

examples/plot_display_object_visualization.py

@@ -0,0 +1,91 @@
+"""
+===================================
+Visualizations with Display Objects


Update the visualization UG to link to this example? (just a link in the example box is enough)

NicolasHug · 2020-04-21T18:30:04Z

examples/plot_display_object_visualization.py

+:class:`ConfusionMatrixDisplay`, :class:`RocCurveDisplay`, and
+:class:`PrecisionRecallDisplay` directly from their respective metrics. This
+is an alternative to using their corresponding plot functions when
+a model's predictions are already computed or expensive to compute.


I'd suggest insisting further that this is advanced usage:

Note that this is advanced usage, and in general we recommend using their respective plot functions.

NicolasHug · 2020-04-21T18:30:39Z

examples/plot_display_object_visualization.py

+# -------------------------
+# For this example, we load a blood transfusion service center data set
+# from `OpenML <https://www.openml.org/d/1464>`. This is a binary
+# classification problem where the target is if an individual donated blood.


Suggested change

# classification problem where the target is if an individual donated blood.

# classification problem where the target is whether an individual donated blood.

not sure

NicolasHug · 2020-04-21T18:31:57Z

examples/plot_display_object_visualization.py

+# dataset. These predictions are used to compute confustion matrix and
+# plot with the :class:`ConfusionMatrixDisplay`


Suggested change

# dataset. These predictions are used to compute confustion matrix and

# plot with the :class:`ConfusionMatrixDisplay`

# dataset. These predictions are used to compute the confusion matrix which

# is plotted with the :class:`ConfusionMatrixDisplay`

NicolasHug · 2020-04-21T18:32:58Z

examples/plot_display_object_visualization.py

+##############################################################################
+# Create :class:`RocCurveDisplay`
+##############################################################################
+# The roc curve requires either the probability or the non-thresholded


Suggested change

# The roc curve requires either the probability or the non-thresholded

# The roc curve requires either the probabilities or the non-thresholded

NicolasHug · 2020-04-21T18:33:06Z

examples/plot_display_object_visualization.py

+##############################################################################
+# The roc curve requires either the probability or the non-thresholded
+# decision values from the estimator. Since the logistic regression provides
+# a decision function, we will use it to plot he roc curve:


Suggested change

# a decision function, we will use it to plot he roc curve:

# a decision function, we will use it to plot the roc curve:

NicolasHug · 2020-04-21T18:34:58Z

examples/plot_display_object_visualization.py

+# The display objects stores the computed values of metrics. This allows
+# for the visualizations to be easliy combinied using matplotlib's API. In
+# the following example, we place the displays next to each other in a
+# row.


Suggested change

# The display objects stores the computed values of metrics. This allows

# for the visualizations to be easliy combinied using matplotlib's API. In

# the following example, we place the displays next to each other in a

# row.

# The display objects store the computed values of metrics. This allows

# for the visualizations to be easily combined using matplotlib's API. In

# the following example, we place the displays next to each other in a

# row.

I would suggest "... store the computed values that were passed as arguments"

adrinjalali · 2020-04-22T08:26:50Z

tests failing

adrinjalali · 2020-04-23T08:05:32Z

I think the only thing left is a link from the UG to the example.

adrinjalali · 2020-04-23T17:09:45Z

I think all your suggestions are addressed now @NicolasHug , could you maybe have another look?

NicolasHug · 2020-04-23T17:11:54Z

thanks @thomasjpfan

thomasjpfan added 3 commits April 15, 2020 11:38

ENH Makes display objects easier to compute

382bd4c

DOC Update example

e8e850e

Merge remote-tracking branch 'upstream/master' into display_object_ex…

fd017a5

…ample

github-actions bot added the module:metrics label Apr 15, 2020

Merge remote-tracking branch 'upstream/master' into display_object_ex…

763310f

…ample

rth approved these changes Apr 19, 2020

View reviewed changes

thomasjpfan added this to the 0.23 milestone Apr 21, 2020

adrinjalali reviewed Apr 21, 2020

View reviewed changes

NicolasHug approved these changes Apr 21, 2020

View reviewed changes

thomasjpfan added 2 commits April 21, 2020 16:34

CLN Address adrin's comments

ca87102

CLN Address nic's comments

f45413b

FIX Bug

894d4f9

DOC Add link to user guide

0ec0e99

NicolasHug merged commit 923b13c into scikit-learn:master Apr 23, 2020

lucyleeow mentioned this pull request Apr 30, 2020

DOC Remove unnecessary comment #17091

Merged

gio8tisu pushed a commit to gio8tisu/scikit-learn that referenced this pull request May 15, 2020

API Adds defaults to Display Objects (scikit-learn#16933)

b9009f6

viclafargue pushed a commit to viclafargue/scikit-learn that referenced this pull request Jun 26, 2020

API Adds defaults to Display Objects (scikit-learn#16933)

db3fb01

		Display labels for plot. If None, display labels are set to 0 to
		`n_classes`.

	# classification problem where the target is if an individual donated blood.
	# classification problem where the target is whether an individual donated blood.

		# dataset. These predictions are used to compute confustion matrix and
		# plot with the :class:`ConfusionMatrixDisplay`

	# The roc curve requires either the probability or the non-thresholded
	# The roc curve requires either the probabilities or the non-thresholded

	# a decision function, we will use it to plot he roc curve:
	# a decision function, we will use it to plot the roc curve:

Uh oh!

Conversation

thomasjpfan commented Apr 15, 2020

Reference Issues/PRs

What does this implement/fix? Explain your changes.

Uh oh!

rth left a comment

Choose a reason for hiding this comment

Uh oh!

adrinjalali left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

NicolasHug left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

adrinjalali commented Apr 22, 2020

Uh oh!

adrinjalali commented Apr 23, 2020

Uh oh!

adrinjalali commented Apr 23, 2020

Uh oh!

NicolasHug commented Apr 23, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants