scikit-learn · lesteve · Jul 2, 2018 · Mar 7, 2017 · Mar 7, 2017 · Mar 7, 2017
diff --git a/doc/whats_new/v0.20.rst b/doc/whats_new/v0.20.rst
@@ -326,6 +326,9 @@ Decomposition and manifold learning
   :class:`manifold.TSNE`. :issue:`10593` and :issue:`10610` by
   `Tom Dupre la Tour`_.
 
+- Support sparse input in :meth:`manifold.Isomap.fit`. :issue:`8554` by
+  :user:`Leland McInnes <lmcinnes>`.
+
 Metrics
 
 - :func:`metrics.roc_auc_score` now supports binary ``y_true`` other than

diff --git a/sklearn/manifold/isomap.py b/sklearn/manifold/isomap.py
@@ -100,7 +100,7 @@ def __init__(self, n_neighbors=5, n_components=2, eigen_solver='auto',
         self.n_jobs = n_jobs
 
     def _fit_transform(self, X):
-        X = check_array(X)
+        X = check_array(X, accept_sparse='csr')
         self.nbrs_ = NearestNeighbors(n_neighbors=self.n_neighbors,
                                       algorithm=self.neighbors_algorithm,
                                       n_jobs=self.n_jobs)

diff --git a/sklearn/manifold/locally_linear.py b/sklearn/manifold/locally_linear.py
@@ -69,10 +69,9 @@ def barycenter_kneighbors_graph(X, n_neighbors, reg=1e-3, n_jobs=1):
 
     Parameters
     ----------
-    X : {array-like, sparse matrix, BallTree, KDTree, NearestNeighbors}
+    X : {array-like, NearestNeighbors}
         Sample data, shape = (n_samples, n_features), in the form of a
-        numpy array, sparse array, precomputed tree, or NearestNeighbors
-        object.
+        numpy array or a NearestNeighbors object.
 
     n_neighbors : int
         Number of neighbors for each sample.
@@ -194,10 +193,9 @@ def locally_linear_embedding(
 
     Parameters
     ----------
-    X : {array-like, sparse matrix, BallTree, KDTree, NearestNeighbors}
+    X : {array-like, NearestNeighbors}
         Sample data, shape = (n_samples, n_features), in the form of a
-        numpy array, sparse array, precomputed tree, or NearestNeighbors
-        object.
+        numpy array or a NearestNeighbors object.
 
     n_neighbors : integer
         number of neighbors to consider for each point.

diff --git a/sklearn/manifold/tests/test_isomap.py b/sklearn/manifold/tests/test_isomap.py
@@ -10,6 +10,8 @@
 from sklearn import preprocessing
 from sklearn.utils.testing import assert_less
 
+from scipy.sparse import rand as sparse_rand
+
 eigen_solvers = ['auto', 'dense', 'arpack']
 path_methods = ['auto', 'FW', 'D']
 
@@ -122,3 +124,15 @@ def test_isomap_clone_bug():
         model.fit(np.random.rand(50, 2))
         assert_equal(model.nbrs_.n_neighbors,
                      n_neighbors)
+
+
+def test_sparse_input():
+    X = sparse_rand(100, 3, density=0.1, format='csr')
+
+    # Should not error
+    for eigen_solver in eigen_solvers:
+        for path_method in path_methods:
+            clf = manifold.Isomap(n_components=2,
+                                  eigen_solver=eigen_solver,
+                                  path_method=path_method)
+            clf.fit(X)