adding learners.py in lieu of MOALearners.py

learners.py

@@ -0,0 +1,170 @@
+# Create the JVM and add the MOA jar to the classpath
+from prepare_jpype import start_jpype
+start_jpype()
+
+from abc import ABC, abstractmethod
+
+# MOA/Java imports
+from moa.core import Utils
+
+class Classifier(ABC):
+    def __init__(self, schema=None, random_seed=1):
+        self.random_seed = random_seed
+        self.schema = schema
+
+    @abstractmethod
+    def __str__(self):
+        pass
+
+    @abstractmethod
+    def train(self, instance):
+        pass
+
+    @abstractmethod
+    def predict(self, instance):
+        pass
+
+    @abstractmethod
+    def predict_proba(self, instance):
+        pass
+
+class MOAClassifier(Classifier):
+    def __init__(self, schema=None, CLI=None, random_seed=1, moa_learner=None):
+        super().__init__(schema=schema, random_seed=random_seed)
+        self.CLI = CLI
+        self.moa_learner = moa_learner
+
+        self.moa_learner.setRandomSeed(self.random_seed)
+
+        if self.schema is not None:
+            self.moa_learner.setModelContext(self.schema.get_moa_header())
+
+        if self.CLI is not None:
+            self.moa_learner.getOptions().setViaCLIString(CLI)
+
+        self.moa_learner.prepareForUse()
+        self.moa_learner.resetLearningImpl()
+
+    def __str__(self):
+        # Remove the package information from the name of the learner. 
+        full_name = str(self.moa_learner.getClass().getCanonicalName())
+        return full_name.rsplit(".", 1)[1] if "." in full_name else full_name
+
+    # def describe(self):
+    #     return str(self.moa_learner)
+
+    def CLI_help(self):
+        return str(self.moa_learner.getOptions().getHelpString())
+
+    def train(self, instance):
+        self.moa_learner.trainOnInstance(instance.get_MOA_InstanceExample())
+
+    def predict(self, instance):
+        return Utils.maxIndex(self.moa_learner.getVotesForInstance(instance.get_MOA_InstanceExample()))
+
+    def predict_proba(self, instance):
+        return self.moa_learner.getVotesForInstance(instance.get_MOA_InstanceExample())
+
+
+class SKClassifier(Classifier):
+    def __init__(self, schema=None, random_seed=1, sklearner=None):
+        super().__init__(schema=schema, random_seed=random_seed)
+        self.sklearner = sklearner
+        self.trained_at_least_once = False
+
+    def __str__(self):
+        return 'sklearner' # TODO: get the string describing the sklearner
+
+    def train(self, instance):
+        self.sklearner.partial_fit([instance.x()], [instance.y()], classes=self.schema.get_label_indexes())
+        self.trained_at_least_once = True # deve (e tem que) ter um jeito melhor
+
+    def predict(self, instance):
+        if self.trained_at_least_once: # scikit-learn does not allows invoking predict in a model that was not fit before
+            return self.sklearner.predict([instance.x()])
+        else:
+            return None
+
+    def predict_proba(self, instance):
+        if self.trained_at_least_once: # scikit-learn does not allows invoking predict in a model that was not fit before
+            return self.sklearner.predict_proba([instance.x()])
+        else:
+            return None
+
+
+##############################################################
+############################# SSL ############################
+##############################################################
+
+class ClassifierSSL(Classifier):
+    def __init__(self, schema=None, random_seed=1):
+        super().__init__(schema=schema, random_seed=random_seed)
+
+    @abstractmethod
+    def train_on_unlabeled(self, instance):
+        pass
+
+# Multiple inheritance
+class MOAClassifierSSL(MOAClassifier, ClassifierSSL):
+    def train_on_unlabeled(self, instance):
+        self.moa_learner.trainOnUnlabeledInstance(instance.get_MOA_InstanceExample().getData())
+
+##############################################################
+######################### REGRESSORS #########################
+##############################################################
+
+class Regressor(ABC):
+    def __init__(self, schema=None, random_seed=1):
+        self.random_seed = random_seed
+        self.schema = schema
+
+    def __str__(self):
+        pass
+
+    @abstractmethod
+    def train(self, instance):
+        pass
+
+    @abstractmethod
+    def predict(self, instance):
+        pass
+
+
+class MOARegressor(Regressor):
+    def __init__(self, schema=None, CLI=None, random_seed=1, moa_learner=None):
+        super().__init__(schema=schema, random_seed=random_seed)
+        self.CLI = CLI
+        self.moa_learner = moa_learner
+
+        self.moa_learner.setRandomSeed(random_seed)
+
+        if self.schema is not None:
+            self.moa_learner.setModelContext(self.schema.get_moa_header())
+
+        if self.CLI is not None:
+            self.moa_learner.getOptions().setViaCLIString(CLI)
+
+        self.moa_learner.prepareForUse()
+        self.moa_learner.resetLearning()
+
+    def __str__(self):
+        full_name = str(self.moa_learner.getClass().getCanonicalName())
+        return full_name.rsplit(".", 1)[1] if "." in full_name else full_name
+
+    # def describe(self):
+    #     return str(self.moa_learner)
+
+    def CLI_help(self):
+        return self.moa_learner.getOptions().getHelpString()
+
+    def train(self, instance):
+        self.moa_learner.trainOnInstance(instance.get_MOA_InstanceExample())
+
+    def predict(self, instance):
+        prediction_array = self.moa_learner.getVotesForInstance(instance.get_MOA_InstanceExample())
+        # The learner didn't provide a prediction, returns 0.0 (probably the learner has not been initialised.)
+        if len(prediction_array) == 0:
+            return 0.0
+        return prediction_array[0]
+
+