Fixing error in CUR test

Author: rosecers

src/skmatter/decomposition/_pcovr.py

@@ -181,7 +181,7 @@ class PCovR(_BasePCA, LinearModel):
     >>> Y = np.array([[0, -5], [-1, 1], [1, -5], [-3, 2]])
     >>> pcovr = PCovR(mixing=0.1, n_components=2)
     >>> pcovr.fit(X, Y)
-    PCovR(mixing=0.1, n_components=2, space='sample')
+    PCovR(mixing=0.1, n_components=2)
     >>> pcovr.transform(X)
     array([[ 3.2630561 ,  0.06663787],
            [-2.69395511, -0.41582771],

tests/test_check_estimators.py

@@ -6,10 +6,7 @@
 from skmatter.feature_selection import PCovCUR as fPCovCUR
 from skmatter.feature_selection import PCovFPS as fPCovFPS
 from skmatter.linear_model import RidgeRegression2FoldCV  # OrthogonalRegression,
-from skmatter.preprocessing import (
-    KernelNormalizer,
-    StandardFlexibleScaler,
-)
+from skmatter.preprocessing import KernelNormalizer, StandardFlexibleScaler
 
 
 @parametrize_with_checks(

tests/test_sample_simple_cur.py

@@ -1,7 +1,6 @@
 import unittest
 
 import numpy as np
-from sklearn import exceptions
 from sklearn.datasets import fetch_california_housing as load
 
 from skmatter.sample_selection import CUR, FPS
@@ -10,12 +9,12 @@
 class TestCUR(unittest.TestCase):
     def setUp(self):
         self.X, _ = load(return_X_y=True)
-        self.X = FPS(n_to_select=100).fit(self.X).transform(self.X)
+        self.X = self.X[FPS(n_to_select=100).fit(self.X).selected_idx_]
         self.n_select = min(20, min(self.X.shape) // 2)
 
     def test_bad_transform(self):
         selector = CUR(n_to_select=2)
-        with self.assertRaises(exceptions.NotFittedError):
+        with self.assertRaises(ValueError):
             _ = selector.transform(self.X)
 
     def test_restart(self):