Update to 1.0.0:

change the dataset.y to be 2-dimensional numpy array for single-target data set.

mgktools/__init__.py

@@ -2,4 +2,4 @@
 # -*- coding: utf-8 -*-
 
 
-__version__ = '0.2.1'
+__version__ = '1.0.0'

mgktools/data/data.py

@@ -405,10 +405,7 @@ def X(self) -> np.ndarray:
     @property
     def y(self):
         y = concatenate([d.targets for d in self.data], axis=0)
-        if y is not None and y.shape[1] == 1:
-            return y.ravel()
-        else:
-            return y
+        return y
 
     @property
     def repr(self) -> np.ndarray:  # 2d array str.

mgktools/evaluators/cross_validation.py

@@ -159,9 +159,13 @@ def evaluate_train_test(self, dataset_train: Dataset,
 
         X_train = dataset_train.X
         y_train = dataset_train.y
+        if y_train.shape[1] == 1:
+            y_train = y_train.ravel()
         repr_train = dataset_train.repr.ravel()
         X_test = dataset_test.X
         y_test = dataset_test.y
+        if y_test.shape[1] == 1:
+            y_test = y_test.ravel()
         repr_test = dataset_test.repr.ravel()
         # Find the most similar sample in training sets.
         if self.n_similar is None:
@@ -187,6 +191,8 @@ def evaluate_train_test(self, dataset_train: Dataset,
 
     def _evaluate_loocv(self):
         X, y, repr = self.dataset.X, self.dataset.y, self.dataset.repr.ravel()
+        if y.shape[1] == 1:
+            y = y.ravel()
         if self.n_similar is not None:
             y_similar = self.get_similar_info(X, X, repr, self.n_similar)
         else:

test/data/test_data_mixture.py

@@ -17,7 +17,7 @@
 
 
 @pytest.mark.parametrize('testset', [
-    (['targets_1'], (4,)),
+    (['targets_1'], (4, 1)),
     (['targets_1', 'targets_2'], (4, 2)),
 ])
 def test_only_graph(testset):
@@ -46,7 +46,7 @@ def test_only_fingerprints(testset):
                               features_generator=[features_generator],
                               features_combination=features_combination)
     assert dataset.X.shape == (4, n_features)
-    assert dataset.y.shape == (4,)
+    assert dataset.y.shape == (4, 1)
 
 
 @pytest.mark.parametrize('testset', [
@@ -66,4 +66,4 @@ def test_graph_fingerprints(testset):
                               features_combination=features_combination)
     dataset.graph_kernel_type = 'graph'
     assert dataset.X.shape == (4, 1 + n_features)
-    assert dataset.y.shape == (4,)
+    assert dataset.y.shape == (4, 1)

test/data/test_data_pure.py

@@ -12,7 +12,7 @@
 
 
 @pytest.mark.parametrize('testset', [
-    (['targets_1'], (4,)),
+    (['targets_1'], (4, 1)),
     (['targets_1', 'targets_2'], (4, 2)),
 ])
 def test_only_graph(testset):
@@ -38,7 +38,7 @@ def test_only_fingerprints(testset):
                               target_columns=['targets_1'],
                               features_generator=[features_generator])
     assert dataset.X.shape == (4, n_features)
-    assert dataset.y.shape == (4,)
+    assert dataset.y.shape == (4, 1)
 
 
 @pytest.mark.parametrize('testset', [
@@ -55,4 +55,4 @@ def test_graph_fingerprints(testset):
                               features_generator=[features_generator])
     dataset.graph_kernel_type = 'graph'
     assert dataset.X.shape == (4, 1 + n_features)
-    assert dataset.y.shape == (4,)
+    assert dataset.y.shape == (4, 1)

test/interpret/test_interpret.py

@@ -27,7 +27,7 @@ def test_interpret_training_mols(testset):
         mgk_hyperparameters_file=mgk_hyperparameters_file,
         n_jobs=6)
     for i, df in enumerate(df_interpret):
-        assert df['contribution_value'].sum() == pytest.approx(y_pred[i], 1e-5)
+        assert df['contribution_value'].sum() == pytest.approx(y_pred[i], 1e-4)
 
 
 @pytest.mark.parametrize('testset', [