8000 sklearn.utils._param_validation._InstancesOf is insufficient for numpy data types · Issue #23599 · scikit-learn/scikit-learn · GitHub
[go: up one dir, main page]
Skip to content

sklearn.utils._param_validation._InstancesOf is insufficient for numpy data types #23599

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.

Sign up for GitHub

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
Closed
Diadochokinetic opened this issue Jun 12, 2022 · 2 comments
Closed

sklearn.utils._param_validation._InstancesOf is insufficient for numpy data types #23599

Diadochokinetic opened this issue Jun 12, 2022 · 2 comments
Labels
Bug Needs Triage Issue requires triage

Comments

@Diadochokinetic
Copy link
Contributor

Describe the bug

Numpy data types can be constructed in different ways. Although they result in the same data type, isinstance() yields different results based on the way the data type has been constructed.

>>> isinstance(np.float64, type)
True
>>> isinstance(np.float64, np.dtype)
False
>>> isinstance(np.dtype('float64'), np.dtype)
True
>>> isinstance(np.dtype('float64'), type)
False
>>>

Therefore, if an object accepts numpy data types as a parameter, the parameter constraints need to be [type, np.dtype]. Currently this isn't supported, because sklearn.utils._param_validation._InstancesOf initalizes with the built-in type

 def __init__(self, type):
        self.type = type

This results in errors when trying to implement parameter_constraints in objects with numpy data types as parameters. E.g. #23579

I therefore propose to change the parameter self.type to self.param_type and initialize it with the actual type given within the list of parameter constraints.

Steps/Code to Reproduce

See #23579

Expected Results

No error is thrown.

Actual Results

ValueError: The 'dtype' parameter of OneHotEncoder must be an instance of 'type'. Got dtype('float64') instead.

Versions

System:
    python: 3.8.13 (default, Mar 28 2022, 11:38:47)  [GCC 7.5.0]
executable: /home/fabian/anaconda3/envs/sklearn-dev/bin/python3
   machine: Linux-5.13.0-48-generic-x86_64-with-glibc2.17

Python dependencies:
      sklearn: 1.2.dev0
          pip: 21.2.4
   setuptools: 61.2.0
        numpy: 1.17.3
        scipy: 1.3.2
       Cython: None
       pandas: None
   matplotlib: None
       joblib: 1.0.0
threadpoolctl: 2.0.0

Built with OpenMP: True

threadpoolctl info:
       filepath: /home/fabian/anaconda3/envs/sklearn-dev/lib/libgomp.so.1.0.0
         prefix: libgomp
       user_api: openmp
   internal_api: openmp
        version: None
    num_threads: 12

       filepath: /home/fabian/anaconda3/envs/sklearn-dev/lib/python3.8/site-packages/numpy/.libs/libopenblasp-r0-34a18dc3.3.7.so
         prefix: libopenblas
       user_api: blas
   internal_api: openblas
        version: 0.3.7
    num_threads: 12
threading_layer: pthreads

       filepath: /home/fabian/anaconda3/envs/sklearn-dev/lib/python3.8/site-packages/scipy/.libs/libopenblasp-r0-2ecf47d5.3.7.dev.so
         prefix: libopenblas
       user_api: blas
   internal_api: openblas
        version: 0.3.7.dev
    num_threads: 12
threading_layer: pthreads
@Diadochokinetic Diadochokinetic added Bug Needs Triage Issue requires triage labels Jun 12, 2022
This was referenced Jun 12, 2022
Closed
Merged
@thomasjpfan
Copy link
Member
thomasjpfan commented Jun 12, 2022 •
edited
Loading

The "type" here is not the builtin type anymore but the locally scoped "type". A simple Python example:

class A:
    def __init__(self, type):
        self.type = type
        
a = A(100)
print(a.type)
# 100

@Diadochokinetic
Copy link
Contributor Author

Oh, now I understand. Sorry for the confusion. I'm going to close this issue and the corresponding PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
Bug Needs Triage Issue requires triage
Projects
None yet
Milestone
No milestone
2 participants
@thomasjpfan @Diadochokinetic
0