TST Extend tests for scipy.sparse.*array in test_lof.py (scikit-learn#27162)

Author: bharatr21

sklearn/neighbors/tests/test_lof.py

@@ -7,7 +7,6 @@
 
 import numpy as np
 import pytest
-from scipy.sparse import csr_matrix
 
 from sklearn import metrics, neighbors
 from sklearn.datasets import load_iris
@@ -18,6 +17,7 @@
     check_outlier_corruption,
     parametrize_with_checks,
 )
+from sklearn.utils.fixes import CSR_CONTAINERS
 
 # load the iris dataset
 # and randomly permute it
@@ -238,11 +238,12 @@ def test_predicted_outlier_number(expected_outliers):
         check_outlier_corruption(num_outliers, expected_outliers, y_dec)
 
 
-def test_sparse():
+@pytest.mark.parametrize("csr_container", CSR_CONTAINERS)
+def test_sparse(csr_container):
     # LocalOutlierFactor must support CSR inputs
     # TODO: compare results on dense and sparse data as proposed in:
     # https://github.com/scikit-learn/scikit-learn/pull/23585#discussion_r968388186
-    X = csr_matrix(iris.data)
+    X = csr_container(iris.data)
 
     lof = neighbors.LocalOutlierFactor(novelty=True)
     lof.fit(X)