Closed
Description
Describe the bug
It seems that in the new version 1.3 f1_score with zero_division=1 incorrectly calculates the results. It treats the prediction with 0 true positives as a division by zero and assigned 1. This was not the case in 1.2.X version and below, which were triggering division by zero only when true positives, false negatives, and false positives were 0. The example below demonstrates the issue. This bug may result in over-positive estimates, especially when using macro averaging in multi-label classification.
Steps/Code to Reproduce
y_true = np.array([0, 1])
y_pred = np.array([1, 0])
print(f1_score(y_true, y_pred, zero_division=1)) # Should be 0.0
y_true = np.array([0, 1])
y_pred = np.array([0, 1])
print(f1_score(y_true, y_pred, zero_division=1)) # Should be 1.0
y_true = np.array([0, 1])
y_pred = np.array([0, 0])
print(f1_score(y_true, y_pred, zero_division=1)) # Should be 0.0
y_true = np.array([0, 0])
y_pred = np.array([0, 0])
print(f1_score(y_true, y_pred, zero_division=1)) # Here division by zero should be triggered resulting in 1.0
Expected Results
In versions 1.2.X and below, the correct results are printed, that is 0.0, 1.0, 0.0, 1.0,
Actual Results
The new version 1.3 prints 1.0, 1.0, 0.0, 1.0 - the first value is incorrect.
Versions
System:
python: 3.10.6 (main, May 29 2023, 11:10:38) [GCC 11.3.0]
executable: /usr/bin/python
machine: Linux-5.19.0-50-generic-x86_64-with-glibc2.35
Python dependencies:
sklearn: 1.3.0
pip: 22.0.2
setuptools: 59.6.0
numpy: 1.25.1
scipy: 1.10.0
Cython: None
pandas: 1.5.3
matplotlib: 3.7.1
joblib: 1.2.0
threadpoolctl: 3.1.0
Built with OpenMP: True
threadpoolctl info:
user_api: openmp
internal_api: openmp
prefix: libgomp
filepath: /home/marek/.local/lib/python3.10/site-packages/scikit_learn.libs/libgomp-a34b3233.so.1.0.0
version: None
num_threads: 16
user_api: blas
internal_api: openblas
prefix: libopenblas
filepath: /usr/local/lib/python3.10/dist-packages/numpy.libs/libopenblas64_p-r0-7a851222.3.23.so
version: 0.3.23
threading_layer: pthreads
architecture: Zen
num_threads: 16
user_api: blas
internal_api: openblas
prefix: libopenblas
filepath: /home/marek/.local/lib/python3.10/site-packages/scipy.libs/libopenblasp-r0-41284840.3.18.so
version: 0.3.18
threading_layer: pthreads
architecture: Zen
num_threads: 16