Represent continuous attributes as a bitvector

Author: markspanbroek

balance.py

@@ -634,7 +634,7 @@ def sample(ins, out):
     sample([5, 5, 5, 2], 0),
     sample([5, 5, 5, 3], 0),
     sample([5, 5, 5, 4], 0),
-    continuous_attributes=[0, 1, 2, 3]
+    continuous=[True, True, True, True]
 )
 
 

src/dataset.py

@@ -44,7 +44,7 @@ async def __output__(self):
 class __Dataset__(ObliviousSequence):
     samples: ObliviousArray
     number_of_attributes: int
-    continuous_attributes: [int]
+    continuous: [bool]
 
     def len(self):
         return self.samples.len()
@@ -80,26 +80,27 @@ def select(self, *include):
         return ObliviousDatasetSelection(
             selection,
             self.number_of_attributes,
-            self.continuous_attributes
+            self.continuous
         )
 
     def is_continuous(self, attribute_index):
-        return attribute_index in self.continuous_attributes
+        return self.continuous[attribute_index]
 
     async def __output__(self):
         return await output(self.samples)
 
 
 class ObliviousDataset(__Dataset__, Secret):
-    def __init__(self, values, continuous_attributes=()):
+    def __init__(self, values, continuous=None):
         number_of_attributes = len(values[0]) if len(values) > 0 else 0
         samples = ObliviousArray.create(values)
-        __Dataset__.__init__(
-            self, samples, number_of_attributes, continuous_attributes)
+        if not continuous:
+            continuous = [False for i in range(number_of_attributes)]
+        __Dataset__.__init__(self, samples, number_of_attributes, continuous)
 
     @classmethod
-    def create(cls, *values, continuous_attributes=()):
-        return ObliviousDataset(values, continuous_attributes)
+    def create(cls, *values, continuous=None):
+        return ObliviousDataset(values, continuous)
 
     def __len__(self):
         return len(self.samples)

tests/best_split.py

@@ -43,7 +43,7 @@ def test_select_best_continuous_attribute(self):
             Sample([s(3)], s(0)),
             Sample([s(4)], s(1)),
             Sample([s(5)], s(1)),
-            continuous_attributes=[0]
+            continuous=[True]
         )
         (best_attribute, threshold) = select_best_attribute(samples)
         self.assertEqual(reveal(best_attribute), 0)
@@ -56,7 +56,7 @@ def test_select_best_attribute_from_continuous_and_binary(self):
             Sample([s(1), s(3)], s(0)),
             Sample([s(1), s(4)], s(1)),
             Sample([s(1), s(5)], s(1)),
-            continuous_attributes=[1]
+            continuous=[False, True]
         )
         (best_attribute, threshold) = select_best_attribute(samples)
         self.assertEqual(reveal(best_attribute), 1)

tests/dataset.py

@@ -94,7 +94,7 @@ def test_continuous_attributes(self):
         dataset = ObliviousDataset.create(
             Sample([s(0), s(1), s(1)], s(0)),
             Sample([s(1), s(2), s(1)], s(1)),
-            continuous_attributes=[1]
+            continuous=[False, True, False]
         )
         self.assertFalse(dataset.is_continuous(0))
         self.assertTrue(dataset.is_continuous(1))

tests/example.py

@@ -42,5 +42,5 @@ def sample(ins, out):
     sample([3, 3, 2, 4], 0),
     sample([3, 3, 2, 5], 1),
     sample([3, 3, 3, 1], 0),
-    continuous_attributes=[0, 1, 2, 3]
+    continuous=[True, True, True, True]
 )

tests/train.py

@@ -60,7 +60,7 @@ def test_continuous_attributes(self):
         samples = ObliviousDataset.create(
             Sample([s(1), s(2)], s(0)),
             Sample([s(1), s(3)], s(1)),
-            continuous_attributes=[1])
+            continuous=[False, True])
         self.assertEqual(
             reveal(train(samples, depth=1)),
             Branch(1, threshold=2, left=leaf(0), right=leaf(1)))