Removed underscore from affine_parameters method

sksfa/_sfa.py

@@ -433,12 +433,14 @@ def is_partially_fitted(self):
         except AttributeError:
             return False
 
-    def affine_parameters_(self):
+    def affine_parameters(self):
         """
-        This function provides the parameters of a trained linear SFA model.  
-        Only works if the standard fit method has been used (no partial_fit) and trivial dimensions are either not filled or only filled in 'fill_mode' == 'zero'.
-        Returns the parameters W, b so that the "np.dot(data, W.T) + b" should be equivalent to calling the 'transform' method.
+        This function provides the parameters of a trained linear SFA model.
+        Only works if the standard 'fit' method has been used (no 'partial_fit') and trivial dimensions are either not filled or only filled in 'fill_mode' == 'zero'.
+        Returns the parameters W, b so that the "np.dot(data, W.T) + b" is equivalent to calling the 'transform' method.
         """
+        if not self.is_fitted:
+            raise AttributeError("Transformer has to be fitted to data before extracting parameters")
         out_dim = self.pca_diff_.components_.shape[0]
         n_missing_components = max(self.n_components_ - out_dim, 0)
         print(n_missing_components)

sksfa/tests/test_sfa.py

@@ -77,7 +77,7 @@ def test_sfa_parameter_computation(dimension, n_samples):
     current_data = mixed_trigonometric_functions(dimension, n_samples)
     sfa = SFA()
     slow_features = sfa.fit_transform(current_data)
-    W, b = sfa.affine_parameters_()
+    W, b = sfa.affine_parameters()
     affine_transformed = np.dot(current_data, W.T) + b
     assert np.allclose(slow_features, affine_transformed)
 
@@ -87,7 +87,7 @@ def test_sfa_parameter_computation_rank_deficit_zero_fill(dimension, rank_defici
     current_data = mixed_trigonometric_functions(dimension, rank_deficit=rank_deficit)
     sfa = SFA(fill_mode="zero")
     slow_features = sfa.fit_transform(current_data)
-    W, b = sfa.affine_parameters_()
+    W, b = sfa.affine_parameters()
     affine_transformed = np.dot(current_data, W.T) + b
     assert np.allclose(slow_features, affine_transformed)
 
@@ -98,4 +98,4 @@ def test_sfa_parameter_computation_rank_deficit_nonzero_fill(dimension, rank_def
     sfa = SFA(fill_mode="noise")
     slow_features = sfa.fit_transform(current_data)
     with pytest.raises(RuntimeError):
-      W, b = sfa.affine_parameters_()
+      W, b = sfa.affine_parameters()