Rank: Skip metrices not supported for sparse data

biolab · Jul 1, 2016 · fa50954 · fa50954
1 parent 4f698ee
commit fa50954
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 0 deletions.
diff --git a/Orange/preprocess/score.py b/Orange/preprocess/score.py
@@ -24,6 +24,7 @@
 class Scorer:
     feature_type = None
     class_type = None
+    supports_sparse_data = None
     preprocessors = [
         RemoveNaNClasses()
     ]
@@ -62,6 +63,8 @@ def score_data(self, data, feature):
 
 
 class SklScorer(Scorer, metaclass=WrapperMeta):
+    supports_sparse_data = True
+
     preprocessors = Scorer.preprocessors + [
         Impute()
     ]
@@ -172,6 +175,7 @@ class ClassificationScorer(Scorer):
     """
     feature_type = DiscreteVariable
     class_type = DiscreteVariable
+    supports_sparse_data = True
     preprocessors = Scorer.preprocessors + [
         Discretize(remove_const=False)
     ]
@@ -302,6 +306,7 @@ class ReliefF(Scorer):
     """
     feature_type = Variable
     class_type = DiscreteVariable
+    supports_sparse_data = False
 
     def __init__(self, n_iterations=50, k_nearest=10):
         self.n_iterations = n_iterations
@@ -324,9 +329,11 @@ def score_data(self, data, feature):
             return weights[0]
         return weights
 
+
 class RReliefF(Scorer):
     feature_type = Variable
     class_type = ContinuousVariable
+    supports_sparse_data = False
 
     def __init__(self, n_iterations=50, k_nearest=50):
         self.n_iterations = n_iterations

diff --git a/Orange/widgets/data/owrank.py b/Orange/widgets/data/owrank.py
@@ -9,6 +9,7 @@
 from collections import namedtuple
 
 import numpy as np
+from scipy.sparse import issparse
 
 from PyQt4 import QtGui
 from PyQt4.QtCore import Qt
@@ -255,6 +256,10 @@ def setData(self, data):
                 self.error(0, "Cannot handle class variable type %r" %
                            type(self.data.domain.class_var).__name__)
 
+            if issparse(self.data.X):   # keep only measures supporting sparse data
+                self.measures = [m for m in self.measures
+                                 if m.score.supports_sparse_data]
+
             self.ranksModel.setRowCount(len(attrs))
             for i, a in enumerate(attrs):
                 if a.is_discrete: