ENH added consistency checks for args of check_scalar function #22027

mlant · 2021-12-20T13:56:02Z

Added two checks on the args val_min, val_max and include_boudaries of the function check_scalar #21948:

if max_val is None and include_boundaries in ("right"):
       raise ValueError(
           "`include_boundaries`='right' with no `max_val` is inconsistent."
       )

if min_val is None and include_boundaries in ("left"):
       raise ValueError(
           "`include_boundaries`='left' with no `min_val` is inconsistent."
       )

Here is the new output of @sply88 short script :
1: include_boundaries=left -> x=1, min_val=1: pass
2: include_boundaries=left -> x=1, max_val=1: ValueError("include_boundaries='left' with no min_val is inconsistent.")
3: include_boundaries=right -> x=1, min_val=1: ValueError("include_boundaries='right' with no max_val is inconsistent.")
4: include_boundaries=right -> x=1, max_val=1: pass
5: include_boundaries=both -> x=1, min_val=1: pass
6: include_boundaries=both -> x=1, max_val=1: pass
7: include_boundaries=neither -> x=1, min_val=1: ValueError('input == 1, must be > 1.')
8: include_boundaries=neither -> x=1, max_val=1: ValueError('input == 1, must be < 1.')

mlant · 2021-12-20T15:22:52Z

Some checks are not successful but I don't understand the error messages.
Can someone help me?

glemaitre · 2021-12-20T16:43:01Z

@mlant I can confirm that this is not related to your PR but an update of setuptools. We are trying to solve the problem in #22028

I will look at your PR but considering that the test regarding the docstring is passing, it should be OK.

sklearn/utils/validation.py

glemaitre · 2021-12-20T16:46:42Z

sklearn/utils/validation.py

+
+    if min_val is None and include_boundaries in ("left"):
+        raise ValueError(
+            "`include_boundaries`='left' with no `min_val` is inconsistent."


Suggested change

"`include_boundaries`='left' with no `min_val` is inconsistent."

"`include_boundaries`='left' without specifying explicitly `min_val` is inconsistent."

glemaitre · 2021-12-20T16:46:57Z

sklearn/utils/validation.py

@@ -1418,6 +1419,16 @@ def check_scalar(
            f"Possible values are: {expected_include_boundaries}."
        )

+    if max_val is None and include_boundaries in ("right"):
+        raise ValueError(
+            "`include_boundaries`='right' with no `max_val` is inconsistent."


Suggested change

"`include_boundaries`='right' with no `max_val` is inconsistent."

"`include_boundaries`='right' without specifying explicitly `max_val` is inconsistent."

mlant · 2021-12-20T18:03:03Z

Thank you for your replies,
I added tests in test_validate.py and modified the messages errors is that alright for you?

mlant · 2021-12-29T10:09:03Z

@glemaitre, could you take a look?

glemaitre

Please add an entry to the change log at doc/whats_new/v1.1.rst. Like the other entries there, please reference this pull request with :pr: and credit yourself (and other contributors if applicable) with :user:.

glemaitre · 2022-01-04T10:45:48Z

sklearn/utils/validation.py

@@ -1418,6 +1419,18 @@ def check_scalar(
            f"Possible values are: {expected_include_boundaries}."
        )

+    if max_val is None and include_boundaries in ("right"):


I assume that here we want to check for both possibilities.

Suggested change

if max_val is None and include_boundaries in ("right"):

if max_val is None and include_boundaries in ("right", "both"):

Then we need to do:

Suggested change

if max_val is None and include_boundaries in ("right"):

if max_val is None and include_boundaries == "right":

glemaitre · 2022-01-04T10:46:04Z

sklearn/utils/validation.py

+            "is inconsistent."
+        )
+
+    if min_val is None and include_boundaries in ("left"):


The same here.

Suggested change

if min_val is None and include_boundaries in ("left"):

if min_val is None and include_boundaries in ("left", "both"):

Suggested change

if min_val is None and include_boundaries in ("left"):

if min_val is None and include_boundaries == "left":

glemaitre · 2022-01-04T10:46:36Z

sklearn/utils/tests/test_validation.py

+            int,
+            2,
+            None,
+            "right",


We should add the case where we also have "both" once it is included in the check below.

mlant · 2022-01-04T22:57:16Z

I found that writing "if min_val is None and include_boundaries in ("left", "both"):" obliges the user to specify 'include_boundary' every time only one boundary is defined, because "both" is the default value. This would be equivalent to removing the default value of this parameter. What's more, it raises dozens of errors in the existing code. Therefore, I don't think it is the behavior we want. This is what I reverted to the previous writing (and only updated the changelog).

glemaitre · 2022-01-05T09:59:29Z

I found that writing "if min_val is None and include_boundaries in ("left", "both"):" obliges the user to specify 'include_boundary' every time only one boundary is defined, because "both" is the default value.

This is a bit annoying but we need "both" to be lenient as a default indeed. We might think about changing the default and make "both" stricter in the future. We can see what a second reviewer will do.

glemaitre · 2022-01-05T09:59:50Z

sklearn/utils/validation.py

@@ -1418,6 +1419,18 @@ def check_scalar(
            f"Possible values are: {expected_include_boundaries}."
        )

+    if max_val is None and include_boundaries in ("right"):


Then we need to do:

Suggested change

if max_val is None and include_boundaries in ("right"):

if max_val is None and include_boundaries == "right":

glemaitre · 2022-01-05T10:00:02Z

sklearn/utils/validation.py

+            "is inconsistent."
+        )
+
+    if min_val is None and include_boundaries in ("left"):


Suggested change

if min_val is None and include_boundaries in ("left"):

if min_val is None and include_boundaries == "left":

glemaitre · 2022-01-05T10:01:07Z

doc/whats_new/v1.1.rst

+- |Enhancement| :func:`check_scalar` raises new errors when inconsistent 
+  boundaries are entered (e.g. `include_boundaries`='left' without
+  specifying explicitly `min_val`). 
+  :pr:`22027` by `Marie Lanternier <mlant>`.


Suggested change

- |Enhancement| :func:`check_scalar` raises new errors when inconsistent

boundaries are entered (e.g. `include_boundaries`='left' without

specifying explicitly `min_val`).

:pr:`22027` by `Marie Lanternier <mlant>`.

- |Fix| :func:`check_scalar` raises an error when `include_boundaries={"left", "right"}`

and the boundaries are not set.

:pr:`22027` by `Marie Lanternier <mlant>`.

glemaitre · 2022-01-05T10:02:01Z

sklearn/utils/validation.py

+        raise ValueError(
+            "`include_boundaries`='left' "
+            "without specifying explicitly `min_val` "
+            "is inconsistent."


Could you wrap the error message on 2 lines only and take profit of the 88 characters width.

glemaitre · 2022-01-05T10:02:07Z

sklearn/utils/validation.py

+        raise ValueError(
+            "`include_boundaries`='right' "
+            "without specifying explicitly `max_val` "
+            "is inconsistent."


Could you wrap the error message on 2 lines only and take profit of the 88 characters width.

mlant · 2022-01-06T08:12:59Z

@glemaitre I have fixed a few typos. Now everything LGTM and is ready for review.

glemaitre

LGTM. Maybe @thomasjpfan wants to have a look.

thomasjpfan

LGTM

scikit-learn#22027)" This reverts commit 1e4ac04.

added consistency checks for args of check_scalar function

c862bfe

github-actions bot added the module:utils label Dec 20, 2021

KiraLazu added 2 commits December 20, 2021 14:58

fix string

5154afe

fix string inconsistnt

e3e2fdc

mlant changed the title ~~added consistency checks for args of check_scalar function~~ ENH added consistency checks for args of check_scalar function Dec 20, 2021

glemaitre reviewed Dec 20, 2021

View reviewed changes

sklearn/utils/validation.py Show resolved Hide resolved

glemaitre reviewed Dec 20, 2021

View reviewed changes

mlant and others added 2 commits December 20, 2021 17:57

Merge branch 'scikit-learn:main' into mlant_21948

3cb4f4c

added tests in test_validation.py and modified errors messages

96b774d

glemaitre reviewed Jan 4, 2022

View reviewed changes

KiraLazu added 5 commits January 4, 2022 18:25

Merge branch 'main' into mlant_21948

87829ee

added 'both' cases in tests and modify changelog

a789643

Merge branch 'main' into mlant_21948

9d1a032

added spaces

a14da43

remove both boundary case

13ccb85

glemaitre reviewed Jan 5, 2022

View reviewed changes

KiraLazu added 3 commits January 5, 2022 14:30

Merge branch 'main' into mlant_21948

1645cb2

modify condition and error message

4fc0461

fix typo

da79e09

glemaitre approved these changes Jan 25, 2022

View reviewed changes

thomasjpfan approved these changes Jan 25, 2022

View reviewed changes

thomasjpfan merged commit 1e4ac04 into scikit-learn:main Jan 25, 2022

thomasjpfan mentioned this pull request Jan 25, 2022

MAINT Fix check_scalar error in trees #22302

Merged

thomasjpfan added a commit to thomasjpfan/scikit-learn that referenced this pull request Jan 26, 2022

Revert "ENH added consistency checks for args of check_scalar function (

6fc2a55

scikit-learn#22027)" This reverts commit 1e4ac04.

This was referenced Mar 13, 2022

Include an informative error message in check_scalar for inconsistent inputs #21948

Closed

Added error checks for invalid combination of include_boundaries and min_value/max_value #22074

Closed

	"`include_boundaries`='left' with no `min_val` is inconsistent."
	"`include_boundaries`='left' without specifying explicitly `min_val` is inconsistent."

	"`include_boundaries`='right' with no `max_val` is inconsistent."
	"`include_boundaries`='right' without specifying explicitly `max_val` is inconsistent."

	if max_val is None and include_boundaries in ("right"):
	if max_val is None and include_boundaries in ("right", "both"):

	if min_val is None and include_boundaries in ("left"):
	if min_val is None and include_boundaries in ("left", "both"):

Uh oh!

ENH added consistency checks for args of check_scalar function #22027

ENH added consistency checks for args of check_scalar function #22027

Uh oh!

Conversation

mlant commented Dec 20, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mlant commented Dec 20, 2021

Uh oh!

glemaitre commented Dec 20, 2021

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mlant commented Dec 20, 2021

Uh oh!

mlant commented Dec 29, 2021

Uh oh!

glemaitre 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!

mlant commented Jan 4, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

glemaitre commented Jan 5, 2022

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!

mlant commented Jan 6, 2022

Uh oh!

glemaitre left a comment

Choose a reason for hiding this comment

Uh oh!

thomasjpfan left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

mlant commented Dec 20, 2021 •

edited

Loading

mlant commented Jan 4, 2022 •

edited

Loading