LinearRegression unknown keyword normalize

[huangtu577]

I just installed the PySINDy library, and it throws an TypeError: LinearRegression.__init__() got an unexpected keyword argument 'normalize' out of the box.
The version that conda installed for scikit-learn without any specification is 1.4.2.
According to the Changelog from scikit-learn , the normalize keyword has been deprecated since version 1.0 and removed since 1.2. However there is a proposed fix in the above linked changelog.

Reproducing code example:

import pysindy
differentiation_method = ps.FiniteDifference(order=2)
feature_library = ps.PolynomialLibrary(degree=3)
optimizer = ps.STLSQ(threshold=0.2)

Error message:

---

TypeError Traceback (most recent call last)
Cell In[15], line 3
1 differentiation_method = ps.FiniteDifference(order=2)
2 feature_library = ps.PolynomialLibrary(degree=3)
----> 3 optimizer = ps.STLSQ(threshold=0.2)

File ~/anaconda3/envs/rebound/lib/python3.10/site-packages/pysindy/optimizers/stlsq.py:105, in STLSQ.init(self, threshold, alpha, max_iter, ridge_kw, normalize, fit_intercept, copy_X, initial_guess)
94 def init(
95 self,
96 threshold=0.1,
(...)
103 initial_guess=None,
104 ):
--> 105 super(STLSQ, self).init(
106 max_iter=max_iter,
107 normalize=normalize,
108 fit_intercept=fit_intercept,
109 copy_X=copy_X,
110 )
112 if threshold < 0:
113 raise ValueError("threshold cannot be negative")

File ~/anaconda3/envs/rebound/lib/python3.10/site-packages/pysindy/optimizers/base.py:81, in BaseOptimizer.init(self, max_iter, normalize, fit_intercept, initial_guess, copy_X)
73 def init(
74 self,
75 max_iter=20,
(...)
79 copy_X=True,
80 ):
---> 81 super(BaseOptimizer, self).init(
82 fit_intercept=fit_intercept, normalize=normalize, copy_X=copy_X
83 )
85 if max_iter <= 0:
86 raise ValueError("max_iter must be positive")

TypeError: LinearRegression.init() got an unexpected keyword argument 'normalize'

PySINDy/Python version information:

1.3.0 3.10.14 (main, May 6 2024, 19:42:50) [GCC 11.2.0]

[Jacob-Stevens-Haas]

Thank you!! This is a great find. I'll get on it soon. Since SINDy.model is a Pipeline, this should be a reasonably easy fix

[ggmirandac]

Hi,

I just wanted to mention that the version that you have on Conda-forge is not up to date, and results on the same issues.

Just that.