@@ -41,3 +41,13 @@ def all_correlations(y, X):
     xs_ += 1e-5  # Handle zeros in x
 
     return (xy - x_ * y_ * n) / n / xs_ / ys_
+
+# If you have scipy installed, then you can compute correlations with
+# scipy.spatial.cdist:
+
+def all_correlations_scipy(y, X):
+    from scipy import spatial
+    y = np.atleast_2d(y)
+    sp = spatial.distance.cdist(X, y, 'correlation')
+    # The "correlation distance" is 1 - corr(x,y); so we invert that to obtain the correlation
+    return 1 - sp.ravel()