[Bug]: spectrochempy.PLSRegression.predict() - not enough values to unpack (expected 2, got 1)

[sschusser]

What happened?

I trained PLSRegression with IR spectra. The loadings look as expected. Unfortunately, runnding predict() results in the error
"ValueError: not enough values to unpack (expected 2, got 1)"

`File ~\Programme (portabel)\WPy64-31090\python-3.10.9.amd64\lib\site-packages\spectrochempy\utils\decorators.py:376, in _set_output.call(self, obj, *args, **kwargs)
374 X_transf.title = self.title
375 # make coordset
--> 376 M, N = X.shape
378 if X_transf.shape == X.shape and self.typex is None and self.typey is None:
379 X_transf.dims = X.dims

ValueError: not enough values to unpack (expected 2, got 1)`

What did you expect to happen?

Getting the predictes result.

Minimal Complete Verifiable Example

No response

Relevant log output

No response

Anything else we need to know?

No response

Environment

INSTALLED VERSIONS

commit: None
python: 3.10.9 (tags/v3.10.9:1dd9be6, Dec 6 2022, 20:01:21) [MSC v.1934 64 bit (AMD64)]
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 140 Stepping 1, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: ('en_US', 'ISO8859-1')

-- dev installation --

FileNotFoundError: [Errno 2] No such file or directory: 'C:\XXXXXXXXXXXXXXX\lib\site-packages\requirements\requirements_dev.txt'

[atravert]

Hello,

Thank you very much for your message. It looks like it is related to the dims attribute of your dataset which corresponds to the names of the dimensions of the NDDataset, usually y and x

Can you check ?, e.g.:

>>> X.dims
['y', 'x'] 

if a single name ['y'], or no name is returned, then it is probably the origin of the error -- you will find more informations here on the dimensions and coordinates:
https://www.spectrochempy.fr/latest/userguide/objects/dataset/dataset.html#

Of course let us know whether this solves your problem.

Thanks,

[sschusser]

I modified my code and programmed PLS Regression by hand from sklearn.cross_decomposition. This works fine.

Checking the dims got me

>>> spectra.dims
['y', 'x']

So, no abnormality in dimensions.

[atravert]

Thank you for your reply. Can you post or email me (arnaud.travert at unicaen.fr) the X and Y datasets with which ou are working -- or minimum sample datasets ? With this in hand I should be able to debug spectrochempy.

If you can't, yes the direct use of sklearn is indeed the solution !

Best,
Arnaud.