[BUG] Cannot run the example code

[tyy4n9]

I just installed pysindy on my computer through "pip install pysindy".
However, I failed to run the example codes. It always shows 'STLSQ' object has no attribute '_preprocess_data'.
I installed python through Anaconda, and I used spyder to run the code.

Reproducing code example:

import numpy as np
import pysindy as ps

t = np.linspace(0, 1, 100)
x = 3 * np.exp(-2 * t)
y = 0.5 * np.exp(t)
X = np.stack((x, y), axis=-1)  # First column is x, second is y

model = ps.SINDy(feature_names=["x", "y"])
model.fit(X, t=t)

Error message:

Traceback (most recent call last):

File "C:\Users\AppData\Local\Temp\ipykernel_11640\1278902783.py", line 16, in <cell line: 16>
model.fit(X, t=t)

File "d:\pysindy.py", line 563, in fit
self.model.fit(x, x_dot)

File "E:\anaconda3\envs\pysindy\lib\site-packages\sklearn\pipeline.py", line 382, in fit
self._final_estimator.fit(Xt, y, **fit_params_last_step)

File "d:\pysindy\optimizers\sindy_optimizer.py", line 58, in fit
self.optimizer.fit(x, y)

File "d:\pysindy\optimizers\base.py", line 134, in fit
x, y, X_offset, y_offset, X_scale = self._preprocess_data(

AttributeError: 'STLSQ' object has no attribute '_preprocess_data'

[akaptano]

What are your pysindy and sklearn versions?

[jcallaham]

I get the same error with pysindy 1.7 and sklearn 1.1.0. Actually about half of the tests seem to be failing. Looks like this might be the culprit:

scikit-learn/scikit-learn#22762

They changed _preprocess_data from a static method to just a regular function. An easy fix would be to import the function from sklearn.linear_models._base. I don't really know if that's good practice, since the function doesn't seem to be intended for public use, but I'm just going to do that for now to get things working again at least.

Also, since _preprocess_data was already a function in that module I think this fix should be backwards compatible.

[tyy4n9]

Thanks for the reply. Yes, I think the problem comes from the version of the scikit-learn. If I created a new env and
installed pysindy on this environment, this will not work (sklearn: 1.0.0). However, if I cloned
the base environment (Anaconda default environment, sklearn: 0.24.1) and installed pysindy on it, this will work.

[akaptano]

@briandesilva Any thoughts on the fix? This is a bit how sklearn version 1.0 broke the column normalization functionality in the sklearn regression.

Thanks Jared for pointing out the issue

[panandreou]

I have the latest version of python and the latest version of pysindy. The error is still occurring. I don't understand what is meant by the newest version of the main branch. pip will not let me install an older version of sklearn.