Add generic test for classify method of concrete openml providers

The method classify was returning opposite value in the feedzai/openml-java project.
In order to improve UTs, the method AbstractProviderModelBaseTest#classifyIndexOfMaxScoresValue() was added in order to test the method #classify for all the providers.

openml-api/src/main/java/com/feedzai/openml/model/ClassificationMLModel.java

@@ -42,7 +42,12 @@ public interface ClassificationMLModel extends MachineLearningModel {
      * @param instance The {@link Instance} to be classified.
      * @return The index of the class nominal value according to the {@link DatasetSchema}
      * provided during training of the model.
+     *
+     * @deprecated The idea is to classify the biggest value of the class probabilities distribution obtained from #getClassDistribution(),
+     *  We no longer need this because we can just obtain the biggest value from the class probabilities distribution itself.
+     *
      */
+    @Deprecated
     int classify(Instance instance);
 
 }

openml-utils/src/test/java/com/feedzai/openml/util/provider/AbstractProviderModelBaseTest.java

@@ -78,6 +78,28 @@ public abstract class AbstractProviderModelBaseTest<M extends ClassificationMLMo
      *                       TESTS                       *
      * * * * * * * * * * * * * * * * * * * * * * * * * * */
 
+
+    /**
+     * Verifies that the {@link ClassificationMLModel#classify(Instance)} " returns the index of the greatest value in
+     * the class probability distribution produced by the calling
+     * {@link ClassificationMLModel#getClassDistribution(Instance)} on the model
+     *
+     * @see ClassificationMLModel
+     */
+    protected void canGetClassDistributionMaxValueIndex(final M model, final Instance instance){
+
+        final double[] scores = model.getClassDistribution(instance);
+
+        final int classificationIndex = model.classify(instance);
+
+        final double maxScore = Arrays.stream(scores).max().getAsDouble();
+
+        assertThat(Arrays.asList(ArrayUtils.toObject(scores)).indexOf(maxScore))
+                .as("The index of maximum value")
+                .isEqualTo(classificationIndex);
+    }
+
+
     /**
      * Checks that is possible to get a {@link MachineLearningProvider} given a valid provider and algorithm.
      */