[BUG]

[wal-iston]

Hi, I don't know if it is a bug.

I am using:
extractor = pymfe.mfe.MFE(features=['f1'], groups=['complexity'], summary=['max']) extractor.fit(np.array(X), y.numpy().astype('int64'))
X is a tensor with dimension (100, 1176).
y is a tensor with dimension (100,).

---

Do you know why am I receiving the following warnings?

/home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/_internal.py:1568: UserWarning: It is not possible make equal discretization
warnings.warn("It is not possible make equal discretization")
/home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/complexity.py:804: RuntimeWarning: invalid value encountered in true_divide
attr_discriminant_ratio = _numer / _denom
/home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/_internal.py:731: RuntimeWarning: Can't summarize feature 'f1' with summary 'max'. Will set it as 'np.nan'.
warnings.warn(

[FelSiq]

It appears that for some features in X the f1 measure is undefined (NaN).

You can use summary=["nanmax"] to ignore the problematic features and compute the maximum among the valid values.

Those warnings, however, hint that something may be off with some features in X. I suggest you to analyze it considering the definition of the f1 measure.

Felipe.

[wal-iston]

Hi, thank you.

What does it mean? " hint that something may be off with some features in X "

---

X has no nan value. Look:

In [54]: np.isnan(X).any() Out[54]: False
The same happens to y.

---

I tried summary=["nanmax"] , but unfortunately I receive the result:

home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/_internal.py:1568: UserWarning: It is not possible make equal discretization warnings.warn("It is not possible make equal discretization") /home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/complexity.py:804: RuntimeWarning: invalid value encountered in true_divide attr_discriminant_ratio = _numer / _denom /home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/_internal.py:1568: UserWarning: It is not possible make equal discretization warnings.warn("It is not possible make equal discretization") /home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/complexity.py:804: RuntimeWarning: invalid value encountered in true_divide attr_discriminant_ratio = _numer / _denom /home/user/miniconda3/envs/complexity/lib/python3.8/site-packages/pymfe/_internal.py:1568: UserWarning: It is not possible make equal discretization warnings.warn("It is not possible make equal discretization")

[FelSiq]

I can't track the cause of those warnings without access to the data.

Can you use any other dataset to check if the issue persists?

[wal-iston]

Indeed, it don't persist with the data from the examples in https://github.com/ealcobaca/pymfe