Merge pull request scikit-multiflow#75 from PGijsbers/fix_40

Set numpy.ndarray as return type in predict() and predict_proba() methods.
- Fixes 40

src/skmultiflow/lazy/sam_knn.py

@@ -376,7 +376,7 @@ def predict(self, X):
         for i in range(r):
             distancesSTM = SAMKNN.get_distances(X[i], self._STMSamples)
             predictedLabel.append(self.predictFct(X[i], None, distancesSTM))
-        return predictedLabel
+        return np.asarray(predictedLabel)
 
     def predict_proba(self, X):
         raise NotImplementedError

src/skmultiflow/meta/adaptive_random_forests.py

@@ -1,4 +1,5 @@
 from copy import deepcopy
+import numpy as np
 from sklearn.preprocessing import normalize
 
 from skmultiflow.core.base_object import BaseObject
@@ -230,7 +231,7 @@ def predict(self, X):
             Samples for which we want to predict the labels.
         Returns
         -------
-        list
+        numpy.ndarray
             Predicted labels for all instances in X.
         """
         r, _ = get_dimensions(X)
@@ -242,7 +243,7 @@ def predict(self, X):
                 predictions.append(0)
             else:
                 predictions.append(max(votes, key=votes.get))
-        return predictions
+        return np.asarray(predictions)
 
     def predict_proba(self, X):
         """ Predicts the class probabilities for the X instance(s).

src/skmultiflow/meta/leverage_bagging.py

@@ -316,8 +316,8 @@ def predict(self, X):
         
         Returns
         -------
-        list
-            A list with the label prediction for all the samples in X.
+        numpy.ndarray
+            A numpy.ndarray with the label prediction for all the samples in X.
         
         """
         r, c = get_dimensions(X)
@@ -328,7 +328,7 @@ def predict(self, X):
             proba = np.zeros((r, 1))
         for i in range(r):
             predictions.append(np.argmax(proba[i]))
-        return predictions
+        return np.asarray(predictions)
 
     def predict_proba(self, X):
         """ predict_proba
@@ -392,7 +392,7 @@ def predict_proba(self, X):
                 aux.append([x / sum_proba[i] for x in proba[i]])
             else:
                 aux.append(proba[i])
-        return aux
+        return np.asarray(aux)
 
     def predict_binary_proba(self, X):
         """ predict_binary_proba

src/skmultiflow/meta/oza_bagging.py

@@ -196,8 +196,8 @@ def predict(self, X):
         
         Returns
         -------
-        list
-            A list with the label prediction for all the samples in X.
+        numpy.ndarray
+            A numpy.ndarray with the label prediction for all the samples in X.
         
         """
         r, c = get_dimensions(X)
@@ -207,7 +207,7 @@ def predict(self, X):
             return None
         for i in range(r):
             predictions.append(np.argmax(proba[i]))
-        return predictions
+        return np.asarray(predictions)
 
     def predict_proba(self, X):
         """ predict_proba
@@ -267,7 +267,7 @@ def predict_proba(self, X):
                 aux.append([x / sum_proba[i] for x in proba[i]])
             else:
                 aux.append(proba[i])
-        return aux
+        return np.asarray(aux)
 
     def score(self, X, y):
         raise NotImplementedError

src/skmultiflow/neural_networks/perceptron.py

@@ -1,3 +1,4 @@
+import numpy as np
 from skmultiflow.core.base import StreamModel
 from sklearn.linear_model.perceptron import Perceptron
 
@@ -107,11 +108,11 @@ def predict(self, X):
 
         Returns
         -------
-        list
-            A list containing the predicted labels for all instances in X.
+        numpy.ndarray
+            A numpy.ndarray containing the predicted labels for all instances in X.
 
         """
-        return self.classifier.predict(X)
+        return np.asarray(self.classifier.predict(X))
 
     def predict_proba(self, X):
         """ predict_proba

src/skmultiflow/trees/regression_hoeffding_tree.py

@@ -555,7 +555,7 @@ def predict(self, X):
 
         Returns
         -------
-        list
+        numpy.ndarray
             Predicted target values.
 
         """
@@ -584,7 +584,7 @@ def predict(self, X):
         else:
             # Model is empty
             predictions.append(0.0)
-        return predictions
+        return np.asarray(predictions)
 
     def predict_proba(self, X):
         pass

tests/bayes/test_naive_bayes.py

@@ -53,4 +53,7 @@ def test_naive_bayes(test_path):
     expected_score = 0.751503006012024
     assert np.isclose(expected_score, learner.score(X=X_batch[4501:], y=y_batch[4501:]))
 
-    assert 'estimator' == learner.get_class_type()
+    assert 'estimator' == learner.get_class_type()
+
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/lazy/test_knn.py

@@ -58,3 +58,6 @@ def test_knn():
     correct_predictions = sum(predictions == y_batch[4501:4550])
     expected_correct_predictions = 49
     assert correct_predictions == expected_correct_predictions
+
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/lazy/test_knn_adwin.py

@@ -52,3 +52,6 @@ def test_knn_adwin():
     correct_predictions = sum(np.array(predictions) == y[951:])
     expected_correct_predictions = 47
     assert correct_predictions == expected_correct_predictions
+
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/lazy/test_sam_knn.py

@@ -41,6 +41,9 @@ def test_sam_knn(package_path):
 
     assert np.alltrue(predictions == expected_predictions)
 
+    assert type(learner.predict(X)) == np.ndarray
+    #  assert type(learner.predict_proba(X)) == np.ndarray  predict_proba not implemented.
+
 
 def test_sam_knn_coverage(package_path):
 

tests/meta/test_adaptive_random_forests.py

@@ -45,6 +45,8 @@ def test_adaptive_random_forests():
         #  Temporary disable as pre-3.6 give different predictions than 3.6+
         assert np.alltrue(predictions == last_version_predictions)
 
+    assert type(learner.predict(X)) == np.ndarray
+
 
 def test_adaptive_random_forests_labels_given():
     stream = RandomTreeGenerator(tree_random_state=112, sample_random_state=112, n_classes=2)
@@ -128,3 +130,5 @@ def test_adaptive_random_forests_batch_predict_proba():
     if sys.version_info.major == 3 and sys.version_info.minor >= 6:
         #  Temporary disable as pre-3.6 give different predictions than 3.6+
         assert np.alltrue(all_predictions.argmax(axis=1) == last_version_predictions)
+
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/meta/test_batch_incremental.py

@@ -48,3 +48,5 @@ def test_batch_incremental():
     assert np.alltrue(predictions == expected_predictions)
     assert np.isclose(expected_performance, performance)
     assert correct_predictions == expected_correct_predictions
+
+    assert type(learner.predict(X)) == np.ndarray

tests/meta/test_classifier_chains.py

@@ -90,6 +90,9 @@ def test_classifier_chains():
     assert np.alltrue(np.array_equal(predictions, expected_predictions))
     assert correct_predictions == expected_correct_predictions
 
+    assert type(learner.predict(X)) == np.ndarray
+    #  assert type(learner.predict_proba(X)) == np.ndarray  Not available because default loss is set to 'hinge'
+
 
 def test_classifier_chains_all():
     seed = 1

tests/meta/test_leverage_bagging.py

@@ -40,3 +40,6 @@ def test_leverage_bagging():
     assert np.alltrue(predictions == expected_predictions)
     assert np.isclose(expected_performance, performance)
     assert correct_predictions == expected_correct_predictions
+
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/meta/test_multi_output_learner.py

@@ -92,4 +92,5 @@ def test_multi_output_learner():
     assert np.isclose(expected_performance, perf)
     assert correct_predictions == expected_correct_predictions
 
-
+    assert type(classifier.predict(X)) == np.ndarray
+    assert type(classifier.predict_proba(X)) == np.ndarray

tests/meta/test_oza_bagging.py

@@ -41,3 +41,5 @@ def test_oza_bagging():
     assert np.isclose(expected_performance, performance)
     assert correct_predictions == expected_correct_predictions
 
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/meta/test_oza_bagging_adwin.py

@@ -41,3 +41,5 @@ def test_oza_bagging_adwin():
     assert np.isclose(expected_performance, performance)
     assert correct_predictions == expected_correct_predictions
 
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/meta/test_regressor_chains.py

@@ -81,8 +81,9 @@ def test_regressor_chains():
                             [-113.86249490223707, 2634310697909.643, 1.580428629322546e+23],
                             [-35.92856878407447, -5410985463428.589, 2.522168862637753e+23]]
 
-
     print(predictions)
     assert np.allclose(np.array(predictions).all(), np.array(expected_predictions).all())
+    assert type(learner.predict(X)) == np.ndarray
+
 
 test_regressor_chains()

tests/neural_networks/test_perceptron.py

@@ -59,3 +59,6 @@ def test_perceptron(test_path):
     # assert np.isclose(expected_accuracy, accuracy)  # Removed due to npn-replicable error in Travis build
 
     assert 'estimator' == learner.get_class_type()
+
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/trees/test_hoeffding_adaptive_tree.py

@@ -59,6 +59,9 @@ def test_hat_mc(test_path):
            or  (learner.get_model_description() == expected_model_4) \
 
 
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray
+
     stream.restart()
     X, y = stream.next_sample(5000)
 
@@ -108,6 +111,8 @@ def test_hat_nb(test_path):
                     ' - nominal_attributes: [] - '
 
     assert learner.get_info() == expected_info
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray
 
 
 def test_hat_nba(test_path):
@@ -150,3 +155,5 @@ def test_hat_nba(test_path):
                     ' - nominal_attributes: [] - '
 
     assert learner.get_info() == expected_info
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/trees/test_hoeffding_tree.py

@@ -53,6 +53,8 @@ def test_hoeffding_tree(test_path):
     expected_model_2 = 'Leaf = Class 1.0 | {1.0: 1745.0, 2.0: 978.0, 0.0: 1423.0, 3.0: 854.0}\n'
     assert (learner.get_model_description() == expected_model_1) \
            or (learner.get_model_description() == expected_model_2)
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray
 
 
 def test_hoeffding_tree_coverage():

tests/trees/test_lc_hoeffding_tree.py

@@ -1,7 +1,6 @@
 import numpy as np
 from skmultiflow.trees import LCHT
 from skmultiflow.data import MultilabelGenerator
-import os
 
 def test_lc_hoeffding_tree(test_path):
     stream = MultilabelGenerator(n_samples=10000, n_features=15, n_targets=3, n_labels=4, random_state=112)
@@ -34,4 +33,6 @@ def test_lc_hoeffding_tree(test_path):
                            [1, 0, 1], [0, 1, 1], [1, 1, 1], [1, 1, 1], [0, 1, 0], [0, 1, 0], [1, 1, 1], [1, 1, 1],
                            [1, 1, 1]]
 
-    assert np.alltrue(predictions == expected_predictions)
+    assert np.alltrue(predictions == expected_predictions)
+    assert type(learner.predict(X)) == np.ndarray
+    assert type(learner.predict_proba(X)) == np.ndarray

tests/trees/test_multi_target_regression_hoeffding_tree.py

@@ -49,6 +49,7 @@ def test_multi_target_regression_hoeffding_tree_mean(test_path):
         'nominal_attributes: [] - '
     assert learner.get_info() == expected_info
     assert isinstance(learner.get_model_description(), type(''))
+    assert type(learner.predict(X)) == np.ndarray
 
 
 def test_multi_target_regression_hoeffding_tree_perceptron(test_path):

tests/trees/test_regression_hoeffding_adaptive_tree.py

@@ -53,6 +53,7 @@ def test_hoeffding_tree():
     assert learner.get_info() == expected_info
 
     assert isinstance(learner.get_model_description(), type(''))
+    assert type(learner.predict(X)) == np.ndarray
 
 
 def test_hoeffding_tree_perceptron():
@@ -103,3 +104,4 @@ def test_hoeffding_tree_perceptron():
     assert learner.get_info() == expected_info
 
     assert isinstance(learner.get_model_description(), type(''))
+    assert type(learner.predict(X)) == np.ndarray

tests/trees/test_regression_hoeffding_tree.py

@@ -54,6 +54,7 @@ def test_hoeffding_tree():
     assert learner.get_info() == expected_info
 
     assert isinstance(learner.get_model_description(), type(''))
+    assert type(learner.predict(X)) == np.ndarray
 
 
 def test_hoeffding_tree_perceptron():
@@ -104,6 +105,7 @@ def test_hoeffding_tree_perceptron():
     assert learner.get_info() == expected_info
 
     assert isinstance(learner.get_model_description(), type(''))
+    assert type(learner.predict(X)) == np.ndarray
 
 
 def test_hoeffding_tree_coverage(test_path):