Closed
Description
Describe the bug
Thank you for the update in #25531
In scikit-learn 1.3.0, there seems to be a bug in the zero_division argument of the f1_score.
It appears to be degrading comparing with scikit-learn 1.2.2.
Steps/Code to Reproduce
>>> from sklearn.metrics import f1_score
>>> f1_score([1,0,0], [0,1,0])
0.0
>>> f1_score([1,0,0], [0,1,0], zero_division=1.0)
1.0Expected Results
Returns 0.0 not 1.0.
In scikit-learn 1.2.2
>>> from sklearn.metrics import f1_score
>>> f1_score([1,0,0], [0,1,0])
0.0
>>> f1_score([1,0,0], [0,1,0], zero_division=1.0)
0.0>>> confusion_matrix([1,0,0], [0,1,0])
array([[1, 1],
[1, 0]])
- precision 0.0
- recall 0.0
- -> f1_score should be 0.0 to show model's low performance
Actual Results
Returns 1.0
Other case (OK)
>>> f1_score([0,0,0], [0,0,0]) # raise UndefinedMetricWarning
0.0
>>> f1_score([0,0,0], [0,0,0], zero_division=1)
1.0Versions
System:
python: 3.9.16 (main, Jan 3 2023, 22:16:38) [Clang 14.0.0 (clang-1400.0.29.202)]
executable: /.../sci13/bin/python
machine: macOS-12.6.6-arm64-arm-64bit
Python dependencies:
sklearn: 1.3.0
pip: 23.2
setuptools: 68.0.0
numpy: 1.25.1
scipy: 1.11.1
Cython: None
pandas: None
matplotlib: None
joblib: 1.3.1
threadpoolctl: 3.2.0
Built with OpenMP: True
threadpoolctl info:
user_api: openmp
internal_api: openmp
num_threads: 10
prefix: libomp
filepath: /.../sci13/lib/python3.9/site-packages/sklearn/.dylibs/libomp.dylib
version: None
user_api: blas
internal_api: openblas
num_threads: 10
prefix: libopenblas
filepath: /.../sci13/lib/python3.9/site-packages/numpy/.dylibs/libopenblas64_.0.dylib
version: 0.3.23
threading_layer: pthreads
architecture: armv8
user_api: blas
internal_api: openblas
num_threads: 10
prefix: libopenblas
filepath: /.../sci13/lib/python3.9/site-packages/scipy/.dylibs/libopenblas.0.dylib
version: 0.3.21.dev
threading_layer: pthreads
architecture: armv8