Closed ftnext closed 1 year ago
Thanks for your wonderful work! I noticed a bug.
mlxtend.evaluate.mcnemar does not return None as chi2 when exact=True.
mlxtend.evaluate.mcnemar
None
chi2
exact=True
ref: https://rasbt.github.io/mlxtend/user_guide/evaluate/mcnemar/#example-3-mcnemars-test-for-scenario-a
>>> import numpy as np >>> from mlxtend.evaluate import mcnemar >>> tb_a = np.array([[9959, 11], [1, 29]]) >>> tb_a array([[9959, 11], [ 1, 29]]) >>> chi2, _ = mcnemar(tb_a, exact=True)
chi2 is None.
ref: https://rasbt.github.io/mlxtend/user_guide/evaluate/mcnemar/#api
if exact=True (default: False), chi2 is None
chi2 is not None.
>>> chi2 1
The value comes from:
https://github.com/rasbt/mlxtend/blob/1943512dbf68ed967ec5a8941d6ce8d96288ff09/mlxtend/evaluate/mcnemar.py#L216
Unit test does not verify that the return value is None.
https://github.com/rasbt/mlxtend/blob/1943512dbf68ed967ec5a8941d6ce8d96288ff09/mlxtend/evaluate/tests/test_mcnemar_test.py#L44
The assertion should be assert chi2p is None
assert chi2p is None
MLxtend 0.21.0 macOS-10.14.6-x86_64-i386-64bit Python 3.10.2 (v3.10.2:a58ebcc701, Jan 13 2022, 14:50:16) [Clang 13.0.0 (clang-1300.0.29.30)] Scikit-learn 1.1.2 NumPy 1.23.3 SciPy 1.9.1
Thanks for your wonderful work! I noticed a bug.
Describe the bug
mlxtend.evaluate.mcnemar
does not returnNone
aschi2
whenexact=True
.Steps/Code to Reproduce
ref: https://rasbt.github.io/mlxtend/user_guide/evaluate/mcnemar/#example-3-mcnemars-test-for-scenario-a
Expected Results
chi2
is None.ref: https://rasbt.github.io/mlxtend/user_guide/evaluate/mcnemar/#api
Actual Results
chi2
is not None.The value comes from:
https://github.com/rasbt/mlxtend/blob/1943512dbf68ed967ec5a8941d6ce8d96288ff09/mlxtend/evaluate/mcnemar.py#L216
Unit test does not verify that the return value is None.
https://github.com/rasbt/mlxtend/blob/1943512dbf68ed967ec5a8941d6ce8d96288ff09/mlxtend/evaluate/tests/test_mcnemar_test.py#L44
The assertion should be
assert chi2p is None
Versions
MLxtend 0.21.0 macOS-10.14.6-x86_64-i386-64bit Python 3.10.2 (v3.10.2:a58ebcc701, Jan 13 2022, 14:50:16) [Clang 13.0.0 (clang-1300.0.29.30)] Scikit-learn 1.1.2 NumPy 1.23.3 SciPy 1.9.1