Matthews correlation coefficient gives wrong results for single-class result #25258
Description
Describe the bug
for a single class input, (ie perfect correlation) MCC returns 0, when it should return 1.
since MCC is a special case of pearson correlation, it should fall back to pearson when appropriate
Steps/Code to Reproduce
from sklearn.metrics import matthews_corrcoef
y_true = [1, 1, 1, 1]
y_pred = [1, 1, 1, 1]
matthews_corrcoef(y_true, y_pred)
Out [9]: 0.0
Expected Results
output should be 1
Actual Results
output is 0
Versions
import sklearn; sklearn.show_versions()
System:
python: 3.9.12 (main, Apr 5 2022, 01:53:17) [Clang 12.0.0 ]
machine: macOS-10.16-x86_64-i386-64bit
Python dependencies:
pip: 22.3.1
setuptools: 65.5.1
sklearn: 1.0.2
numpy: 1.22.3
scipy: 1.9.3
Cython: 0.29.32
pandas: 1.5.2
matplotlib: 3.6.2
joblib: 1.2.0
threadpoolctl: 3.1.0
Built with OpenMP: True
/Users/phild1/opt/anaconda3/lib/python3.9/site-packages/_distutils_hack/__init__.py:33: UserWarning: Setuptools is replacing distutils.
warnings.warn("Setuptools is replacing distutils.")