Why is participation ratio rounded to an integer?

[RylanSchaeffer]

I just noticed that participation ratio is always an integer. Why is that?

scikit-dimension/skdim/id/_PCA.py

Line 189 in ea1a495

de = int(PR)

[j-bac]

Yes it is rounded. If you think it is more useful or standard not to round I can make the change

[RylanSchaeffer]

I've never heard of it being rounded, nor had two of my labmates. We also tried to find whether anyone recommends rounding it in a paper, but weren't successful.

Overall, it's not a big deal. We were just perplexed why we were getting integer values and thought we were misusing the library. Maybe change it?

[j-bac]

Just updated, thanks for the suggestion. Will release on pypi after some other updates

[RylanSchaeffer]

@j-bac any updates on the new release?

[j-bac]

my bad, I did make a new release but seems it didn't include the change. Just added v0.3.1 on pypi

[RylanSchaeffer]

I'm afraid that change introduced an error. Because de is now now longer an integer, this indexing breaks:

scikit-dimension/skdim/id/_PCA.py

Line 192 in ea1a495

return de, gaps[de - 1]

[RylanSchaeffer]

The error:

python3.7/site-packages/skdim/id/_PCA.py", line 192, in _participation_ratio
    return de, gaps[de - 1]
IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

[j-bac]

Thanks, just fixed this (0.3.2). I'm returning all gaps now under self.gap_