Passing Bablok failure

[thomasburgess]

Using methcomp.__version__==1.0.0 and the following data I get totally wrong results with passingbablok, while deming works just fine.

to reproduce the issue:

y1=[-0.44126808, -1.52399143, 2.2107071, -1.00123478, 0.90941934, 1.00586351, 3.68821694, 3.9743334, 2.14226847, 2.64484033, 1.94470087, 4.44356809, 2.38670489, 3.48678895, 3.53981765, 5.23335605, 4.98509282, 4.15024799, 5.57382203, 6.35344638, 5.6439422, 3.54574785, 6.14969295, 6.12347165, 7.39851794, 8.37739543, 9.76299112, 8.01947884, 9.83009684, 6.7872504, 7.84448408, 8.08531552, 9.58223637, 9.86896032, 10.79491967, 10.87242277, 8.69007068, 8.32372217, 12.50016372, 7.90690056, 9.88332462, 10.62266031, 11.18820036, 12.28450113, 12.21681584, 11.96688418, 11.00883212, 13.00745316, 11.97546512, 11.82770948]
y2=[0.46092428, -0.9685343, 1.00745662, 2.89120536, 2.36874212, 2.18474957, 1.25696873, 1.39890868, 3.89918595, 2.4638734, 3.3077234, 2.56369237, 3.54427403, 5.7307152, 4.59708564, 4.65938142, 6.69163919, 4.23384116, 5.82052438, 5.34474272, 6.72508454, 6.04676834, 8.03619467, 8.08816926, 6.4282492, 7.6604346, 8.79697169, 7.34703344, 7.65379082, 9.12290271, 9.55150831, 8.41438351, 7.46967989, 9.25156702, 8.97963457, 10.98358866, 10.07983973, 11.17451036, 10.91000893, 11.75211375, 12.09069388, 12.70345655, 13.68576083, 12.36041789, 13.53545636, 11.52756598, 13.3008177, 15.15793254, 13.46685531, 13.86076108]
fig, (ax1, ax2) = plt.subplots(1,2)
passingbablok(y1, y2, ax=ax1)
deming(y1, y2, ax=ax2)
fig.savefig("catastophic_failure.png")

Here is the resulting graph:

I thought it was the negatives that casued this, but as far as I can tell they do not matter. Any idea what is going wrong?

[thomasburgess]

Note: removing just the first sample "cures" the problem.

[wptmdoorn]

Hi @thomasburgess, thank you for submitting your issue.

I just ran your code with version 1.0.0 and with the most recent master version. The master version produces the correct plots as we would expect:

This is because of a bug in the code, which was found by an user and accordingly fixed in commit https://github.com/wptmdoorn/methcomp/commit/d432fcfdcf1930d71189be24059831c7231855c7. Could you please use the latest version of methcomp to try if this resolves your problem? You can use pip install combined with git to install the master version.

pip install git+https://github.com/wptmdoorn/methcomp

I hope everything is clear for you - please let me know in case you face any other problems or when this does not resolve your problem.

Thanks a lot,

William

[thomasburgess]

Hm, still failing, but my setup is complicated, I'll scratch and redo tomorrow and get back to you if I can't get it working. Now I know where to look too. Thanks!

[wptmdoorn]

Thanks, please let me know if that works - otherwise I will look into it with you :-)

[thomasburgess]

Ok, I think i just failed reinstalling. It now works. I ran it with 3400 random experiments and have no further failures. Something is up with imports now though import methcomp fails, but from methcomp.regression import _PassingBablok as passing_bablok_noplot, deming, passingbablok works fine

[wptmdoorn]

Great, could you give me the full stack trace of the import methcomp statement? That would be very helpful in order to fix the issue.

[thomasburgess]

of course! import methcomp

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-28-4f9017ae7428> in <module>
----> 1 import methcomp

~/.pyenv/versions/3.8.6/envs/jupyter/lib/python3.8/site-packages/methcomp/__init__.py in <module>
      7 from .blandaltman import *
      8 from .regression import *
----> 9 from .glucose import *
     10 
     11 # Define version

~/.pyenv/versions/3.8.6/envs/jupyter/lib/python3.8/site-packages/methcomp/glucose.py in <module>
     18         return pkg_resources.files(pkg).joinpath(file)
     19 
---> 20 from . import static
     21 
     22 __all__ = ["clarke", "parkes", "seg",

ImportError: cannot import name 'static' from partially initialized module 'methcomp' (most likely due to a circular import) (/Users/thomas/.pyenv/versions/3.8.6/envs/jupyter/lib/python3.8/site-packages/methcomp/__init__.py)

[wptmdoorn]

Great find, it's because methcomp.glucose imports the static directory on a module level. I have resolved that issue now and I will push a fix tomorrow to the GitHub. Thank you so much.

[wptmdoorn]

Could you please check if the import errors are resolved now with the master branch of the package?

[thomasburgess]

It is now working!

On Wed, May 5, 2021 at 8:28 PM William van Doorn @.***> wrote:

Could you please check if the import errors are resolved now with the master branch of the package?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/wptmdoorn/methcomp/issues/4#issuecomment-832912883, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAS3HIXQAFV74ZYAKKX7KTTTMGE5RANCNFSM44C7FQLA .

-- Thomas Burgess

[wptmdoorn]

Thanks, will close the issue for now.