def test_target_values(self):
counter = Counter(lambda x: x % 2 == 0)
for i in xrange(25):
counter.add(i)
self.assertEqual(counter[0], 12)
self.assertEqual(counter[1], 13)
counter = Counter(lambda x: None)
for i in xrange(50):
counter.add(i)
self.assertEqual(counter[None], 50)
def _single_node_tree(self):
c = Counter(self.target)
for example in self.dataset:
c.add(example)
node = DecisionTreeNode()
node.set_results_from_counts(c)
return node
def plurality_value(self, examples):
if not examples:
raise ValueError("Dataset is empty")
counter = Counter(self.target)
for example in examples:
counter.add(example)
tree = DecisionTreeNode()
# Note that tie is *not* solved randomly here
tree.set_results_from_counts(counter)
return tree
def classify(self, example):
distances = [(self.problem.distance(e, example), e)
for e in self.dataset]
best = sorted(distances)[:self.k]
counter = Counter(self.problem.target)
for _, example in best:
counter.add(example)
items = [(x[1], x[0]) for x in counter.iteritems()]
items.sort(reverse=True)
return (items[0][1], items[0][0] / counter.total)
def test_add_elements(self):
counter = Counter(lambda x: None)
for i in xrange(20):
counter.add("something")
self.assertEqual(counter.total, 20)
def test_total_starts_in_zero(self):
counter = Counter(lambda x: None)
self.assertEqual(counter.total, 0)
