From 37f25e1f53bde069c1ca6ecf8b32c8ec312440f3 Mon Sep 17 00:00:00 2001
From: Henry Tullis <jhenrytullis@gmail.com>
Date: Fri, 20 Jun 2025 22:50:46 -0700
Subject: [PATCH] compat(sklearn): address _validate_data depr.

make Stabl more compatible with more recent versions of scikit-learn

add backwards-compatible fix for warnings/errors related to the deprication (1.6.0) and removal (1.7.0) of the scikit-learn BaseEstimator._validate_data function

tested this fix using the 'Tutorial Notebook.ipynb' using with various versions of Scikit-learn:
- 1.7.0: runs better than without the fix, however there are still errors and warnings unrelated to this fix
- 1.6.0: runs as expected and reduces warnings
- 1.3.2 (recommended for Stabl): runs as expected

code written with assistance from Claude 3.7
---
 stabl/preprocessing.py | 21 +++++++++++++++++++++
 stabl/stabl.py         | 21 +++++++++++++++++++++
 2 files changed, 42 insertions(+)

diff --git a/stabl/preprocessing.py b/stabl/preprocessing.py
index e536b3de..c74c1eb2 100644
--- a/stabl/preprocessing.py
+++ b/stabl/preprocessing.py
@@ -1,7 +1,10 @@
+from packaging import version
+
 import numpy as np
 from sklearn.base import BaseEstimator
 from sklearn.feature_selection import SelectorMixin
 from sklearn.utils.validation import check_is_fitted
+from sklearn import __version__ as sklearn_version
 
 
 def remove_low_info_samples(X, threshold=1.0):
@@ -131,3 +134,21 @@ def _more_tags(self):
         # Useful to allow the use of nan values
         # For the transform function ;)
         return {"allow_nan": True}
+    
+    def _validate_data(self, *args, **kwargs):
+        """Data validation method compatible with multiple versions of sklearn.
+        
+        For sklearn < 1.6.0, overloads BaseEstimator._validate_data, which is depricated in 1.6.0
+        and removed in 1.7.0. For sklearn >= 1.6.0, this method calls sklearn.utils.validation.validate_data,
+        which performs the equivalent functionality."""
+        
+        skl_ver = version.parse(sklearn_version)
+
+        if skl_ver >= version.parse("1.6.0"):
+            # for scikit-learn >= 1.6.0
+            from sklearn.utils.validation import validate_data
+            return validate_data(self, *args, **kwargs)
+        
+        else:
+            # for scikit-learn < 1.6.0
+            return BaseEstimator._validate_data(self, *args, **kwargs)
diff --git a/stabl/stabl.py b/stabl/stabl.py
index ac111ecd..80ea9925 100644
--- a/stabl/stabl.py
+++ b/stabl/stabl.py
@@ -2,12 +2,14 @@
 from pathlib import Path
 from warnings import warn
 import sys
+from packaging import version
 
 import matplotlib.pyplot as plt
 import numpy as np
 import pandas as pd
 from joblib import Parallel, delayed
 from knockpy.knockoffs import GaussianSampler
+from sklearn import __version__ as sklearn_version
 from sklearn.base import BaseEstimator, clone
 from sklearn.feature_selection import SelectorMixin, SelectFromModel
 from sklearn.linear_model import LogisticRegression, Lasso, ElasticNet
@@ -1064,6 +1066,25 @@ def _validate_input(self):
                 f"When injecting noise, the noise proportion must be between 0 and 1, "
                 f"got {self.artificial_proportion}"
             )
+        
+    def _validate_data(self, *args, **kwargs):
+        """Data validation method compatible with multiple versions of sklearn.
+        
+        For sklearn < 1.6.0, overloads BaseEstimator._validate_data, which is depricated in 1.6.0
+        and removed in 1.7.0. For sklearn >= 1.6.0, this method calls sklearn.utils.validation.validate_data,
+        which performs the equivalent functionality."""
+        
+        skl_ver = version.parse(sklearn_version)
+
+        if skl_ver >= version.parse("1.6.0"):
+            # for scikit-learn >= 1.6.0
+            from sklearn.utils.validation import validate_data
+            return validate_data(self, *args, **kwargs)
+        
+        else:
+            # for scikit-learn < 1.6.0
+            return BaseEstimator._validate_data(self, *args, **kwargs)
+        
 
     def _make_groups(self, X):
         """Make groups for self configuration.