Merge pull request #898 from kernc/Preprocessors,unite!

Unite preprocessor handling in Learner-outputting widgets

Orange/widgets/classify/owclassificationtree.py

@@ -1,19 +1,18 @@
 from Orange.data import Table
-from Orange.preprocess.preprocess import Preprocess
 from Orange.classification.tree import TreeLearner, TreeClassifier
 from Orange.widgets import widget, gui
 from Orange.widgets.settings import Setting
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 from Orange.widgets.utils.sql import check_sql_input
 
 
-class OWClassificationTree(widget.OWWidget):
+class OWClassificationTree(OWProvidesLearner, widget.OWWidget):
     name = "Classification Tree"
     icon = "icons/ClassificationTree.svg"
     description = "Classification tree algorithm with forward pruning."
     priority = 30
 
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
 
     outputs = [
         ("Learner", TreeLearner),
@@ -23,6 +22,7 @@ class OWClassificationTree(widget.OWWidget):
     resizing_enabled = False
 
     LEARNER = TreeLearner
+
     model_name = Setting("Classification Tree")
     attribute_score = Setting(0)
     limit_min_leaf = Setting(True)
@@ -63,10 +63,10 @@ def __init__(self):
         box.layout().addWidget(self.report_button)
         gui.separator(box, 20)
         self.btn_apply = gui.button(box, self, "&Apply",
-                                    callback=self.set_learner, disabled=0,
+                                    callback=self.apply, disabled=0,
                                     default=True)
 
-        self.set_learner()
+        self.apply()
 
     def send_report(self):
         from Orange.canvas.report import plural_w
@@ -85,7 +85,7 @@ def send_report(self):
         if self.data:
             self.report_data("Data", self.data)
 
-    def set_learner(self):
+    def apply(self):
         self.learner = self.LEARNER(
             criterion=self.scores[self.attribute_score][1],
             max_depth=self.max_depth if self.limit_depth else None,
@@ -116,14 +116,7 @@ def set_data(self, data):
         if data is not None and data.domain.class_var is None:
             self.error(0, "Data has no target variable")
             self.data = None
-        self.set_learner()
-
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.set_learner()
+        self.apply()
 
 
 if __name__ == "__main__":

Orange/widgets/classify/owknn.py

@@ -1,17 +1,16 @@
 from Orange.data import Table
 from Orange.classification import KNNLearner, SklModel
-from Orange.preprocess.preprocess import Preprocess
 from Orange.widgets import widget, gui
 from Orange.widgets.settings import Setting
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 from Orange.widgets.utils.sql import check_sql_input
 
 
-class OWKNNLearner(widget.OWWidget):
+class OWKNNLearner(OWProvidesLearner, widget.OWWidget):
     name = "Nearest Neighbors"
     description = "k-nearest neighbors classification algorithm."
     icon = "icons/KNN.svg"
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", KNNLearner), ("Classifier", SklModel)]
 
     want_main_area = False
@@ -50,15 +49,10 @@ def set_data(self, data):
         if data is not None:
             self.apply()
 
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
+    LEARNER = KNNLearner
 
     def apply(self):
-        learner = KNNLearner(
+        learner = self.LEARNER(
             n_neighbors=self.n_neighbors,
             metric=self.metrics[self.metric_index],
             preprocessors=self.preprocessors

Orange/widgets/classify/owlogisticregression.py

@@ -5,19 +5,18 @@
 
 from Orange.data import Table
 from Orange.classification import logistic_regression as lr
-from Orange.preprocess.preprocess import Preprocess
 from Orange.widgets import widget, settings, gui
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 from Orange.widgets.utils.sql import check_sql_input
 
 
-class OWLogisticRegression(widget.OWWidget):
+class OWLogisticRegression(OWProvidesLearner, widget.OWWidget):
     name = "Logistic Regression"
     description = "Logistic regression classification algorithm with " \
                   "LASSO (L1) or ridge (L2) regularization."
     icon = "icons/LogisticRegression.svg"
 
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", lr.LogisticRegressionLearner),
                ("Classifier", lr.LogisticRegressionClassifier)]
 
@@ -85,16 +84,11 @@ def set_data(self, data):
         if data is not None:
             self.apply()
 
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
+    LEARNER = lr.LogisticRegressionLearner
 
     def apply(self):
         penalty = ["l1", "l2"][self.penalty_type]
-        learner = lr.LogisticRegressionLearner(
+        learner = self.LEARNER(
             penalty=penalty,
             dual=self.dual,
             tol=self.tol,

Orange/widgets/classify/owmajority.py

@@ -1,19 +1,18 @@
 from Orange.data import Table
 from Orange.classification.majority import MajorityLearner, ConstantModel
-from Orange.preprocess.preprocess import Preprocess
 from Orange.widgets import widget, gui
 from Orange.widgets.settings import Setting
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 
 
-class OWMajority(widget.OWWidget):
+class OWMajority(OWProvidesLearner, widget.OWWidget):
     name = "Majority"
     description = "Classification to the most frequent class " \
                   "from the training set."
     priority = 20
     icon = "icons/Majority.svg"
 
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", MajorityLearner),
                ("Classifier", ConstantModel)]
 
@@ -44,15 +43,10 @@ def set_data(self, data):
         if data is not None:
             self.apply()
 
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
+    LEARNER = MajorityLearner
 
     def apply(self):
-        learner = MajorityLearner(
+        learner = self.LEARNER(
             preprocessors=self.preprocessors
         )
         learner.name = self.learner_name

Orange/widgets/classify/ownaivebayes.py

@@ -5,16 +5,15 @@
 
 from  Orange.data import Table
 from Orange.classification.naive_bayes import NaiveBayesLearner, NaiveBayesModel
-from Orange.preprocess.preprocess import Preprocess
 from Orange.widgets import widget, gui, settings
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 
 
-class OWNaiveBayes(widget.OWWidget):
+class OWNaiveBayes(OWProvidesLearner, widget.OWWidget):
     name = "Naive Bayes"
     description = "Naive Bayesian classifier."
     icon = "icons/NaiveBayes.svg"
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", NaiveBayesLearner),
                ("Classifier", NaiveBayesModel)]
 
@@ -56,15 +55,10 @@ def set_data(self, data):
         else:
             self.send("Classifier", None)
 
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
+    LEARNER = NaiveBayesLearner
 
     def apply(self):
-        learner = NaiveBayesLearner(
+        learner = self.LEARNER(
             preprocessors=self.preprocessors
         )
         learner.name = self.learner_name

Orange/widgets/classify/owrandomforest.py

@@ -5,27 +5,27 @@
 from PyQt4.QtCore import Qt
 
 from Orange.data import Table
-from Orange.preprocess.preprocess import Preprocess
 from Orange.classification.random_forest import (RandomForestLearner,
                                                  RandomForestClassifier)
 from Orange.widgets import widget, settings, gui
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 from Orange.widgets.utils.sql import check_sql_input
 
 
-class OWRandomForest(widget.OWWidget):
+class OWRandomForest(OWProvidesLearner, widget.OWWidget):
     name = "Random Forest Classification"
     description = "Random forest classification algorithm."
     icon = "icons/RandomForest.svg"
 
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", RandomForestLearner),
                ("Model", RandomForestClassifier)]
 
     want_main_area = False
     resizing_enabled = False
 
     LEARNER = RandomForestLearner
+
     learner_name = settings.Setting("RF Classification Learner")
     n_estimators = settings.Setting(10)
     max_features = settings.Setting(5)
@@ -135,13 +135,6 @@ def set_data(self, data):
         if data is not None:
             self.apply()
 
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
-
     def apply(self):
         common_args = dict()
         common_args["n_estimators"] = self.n_estimators

Orange/widgets/classify/owsvmclassification.py

@@ -7,19 +7,18 @@
 
 from Orange.data import Table
 from Orange.classification.svm import SVMLearner, SVMClassifier, NuSVMLearner
-from Orange.preprocess.preprocess import Preprocess
 from Orange.widgets import widget, settings, gui
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 from Orange.widgets.utils.sql import check_sql_input
 
 
-class OWSVMClassification(widget.OWWidget):
+class OWSVMClassification(OWProvidesLearner, widget.OWWidget):
     name = "SVM"
     description = "Support vector machines classifier with standard " \
                   "selection of kernels."
     icon = "icons/SVM.svg"
 
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", SVMLearner, widget.Default),
                ("Classifier", SVMClassifier),
                ("Support vectors", Table)]
@@ -132,12 +131,7 @@ def set_data(self, data):
         if data is not None:
             self.apply()
 
-    def set_preprocessor(self, preproc):
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
+    LEARNER = SVMLearner  # OWProvidesLearner uses this
 
     def apply(self):
         kernel = ["linear", "poly", "rbf", "sigmoid"][self.kernel_type]

Orange/widgets/regression/owknnregression.py

@@ -4,21 +4,20 @@
 from Orange.data import Table
 from  Orange.regression.knn import KNNRegressionLearner
 from Orange.regression import SklModel
-from Orange.preprocess.preprocess import Preprocess
 
 from Orange.widgets import widget, gui
 from Orange.widgets.settings import Setting
+from Orange.widgets.utils.owlearnerwidget import OWProvidesLearner
 from Orange.widgets.utils.sql import check_sql_input
 
 
-class OWKNNRegression(widget.OWWidget):
+class OWKNNRegression(OWProvidesLearner, widget.OWWidget):
     name = "Nearest Neighbors"
     description = "k-nearest neighbours regression algorithm."
     icon = "icons/kNearestNeighbours.svg"
     priority = 20
 
-    inputs = [("Data", Table, "set_data"),
-              ("Preprocessor", Preprocess, "set_preprocessor")]
+    inputs = [("Data", Table, "set_data")] + OWProvidesLearner.inputs
     outputs = [("Learner", KNNRegressionLearner),
                ("Predictor", SklModel)]
 
@@ -63,19 +62,13 @@ def set_data(self, data):
         if data is not None:
             self.apply()
 
-    def set_preprocessor(self, preproc):
-        """Set preprocessor to apply on training data."""
-        if preproc is None:
-            self.preprocessors = None
-        else:
-            self.preprocessors = (preproc,)
-        self.apply()
+    LEARNER = KNNRegressionLearner
 
     def apply(self):
         """
         Construct the learner and apply it on the training data if available.
         """
-        learner = KNNRegressionLearner(
+        learner = self.LEARNER(
             n_neighbors=self.n_neighbors,
             metric=self.metrics[self.metric_index],
             preprocessors=self.preprocessors