diff --git a/pygsp/graphs/graph.py b/pygsp/graphs/graph.py
index c6fd5397..d63feca3 100644
--- a/pygsp/graphs/graph.py
+++ b/pygsp/graphs/graph.py
@@ -568,20 +568,17 @@ def create_incidence_matrix(self):
         For convenience, the heads and tails of each edge are saved in two
         additional attributes start_nodes and end_nodes.
         """
-        if not hasattr(self, 'directed'):
-            self.is_directed()
-
-        if self.directed or not self.connected:
+        if self.is_directed() or not self.is_connected():
             raise NotImplementedError('Focusing on connected non directed graphs first.')
 
         start_nodes, end_nodes, weights = sparse.find(sparse.tril(self.W))
 
-        data = np.concatenate([np.ones(self.Ne/2), -np.ones(self.Ne/2)])
-        row = np.concatenate([np.arange(self.Ne/2), np.arange(self.Ne/2)])
+        data = np.concatenate([np.ones_like(start_nodes), -np.ones_like(end_nodes)])
+        row = np.concatenate([np.arange(len(start_nodes)), np.arange(len(end_nodes))])
         col = np.concatenate([start_nodes, end_nodes])
 
         self.B = sparse.coo_matrix((data, (row, col)),
-                                   shape=(self.Ne/2, self.N) ).tocsc()
+                                   shape=(len(start_nodes), self.N) ).tocsc()
         self.Wb = sparse.diags(weights,0)
         self.start_nodes = start_nodes
         self.end_nodes = end_nodes
diff --git a/pygsp/reduction.py b/pygsp/reduction.py
index 294d1730..45c4421d 100644
--- a/pygsp/reduction.py
+++ b/pygsp/reduction.py
@@ -339,7 +339,6 @@ def kron_reduction(G, ind, threshold=np.spacing(1)):
 
     """
     if isinstance(G, graphs.Graph):
-
         if G.lap_type != 'combinatorial':
                 msg = 'Unknown reduction for {} Laplacian.'.format(G.lap_type)
                 raise NotImplementedError(msg)
@@ -353,14 +352,13 @@ def kron_reduction(G, ind, threshold=np.spacing(1)):
     else:
 
         L = G
-
     N = np.shape(L)[0]
     ind_comp = np.setdiff1d(np.arange(N, dtype=int), ind)
 
-    L_red = extract_submatrix(L,ind, ind)
-    L_in_out = extract_submatrix(L, ind, ind_comp)
+    L_red = utils.extract_submatrix(L,ind, ind)
+    L_in_out = utils.extract_submatrix(L, ind, ind_comp)
     L_out_in = L_in_out.transpose().tocsc()
-    L_comp = extract_submatrix(L,ind_comp, ind_comp).tocsc()
+    L_comp = utils.extract_submatrix(L,ind_comp, ind_comp).tocsc()
 
     Lnew = L_red - L_in_out.dot(utils.splu_inv_dot(L_comp, L_out_in))