def test_separate_by_attribute(self):
fxnGen = build_instance_generator(0.5)
listInst = fxnGen(self.cInsts)
for ixAttr in xrange(fxnGen.cAttrs):
dictInst = dtree.separate_by_attribute(listInst, ixAttr)
setValues = set([inst.listAttrs[ixAttr] for inst in listInst])
self.assertEqual(len(setValues), len(dictInst))
for cValue,listInstSeparate in dictInst.iteritems():
for inst in listInstSeparate:
self.assertEqual(cValue, inst.listAttrs[ixAttr])
def test_separate_by_attribute(self):
fxnGen = build_instance_generator(0.5)
listInst = fxnGen(self.cInsts)
for ixAttr in xrange(fxnGen.cAttrs):
dictInst = dtree.separate_by_attribute(listInst, ixAttr)
setValues = set([inst.listAttrs[ixAttr] for inst in listInst])
self.assertEqual(len(setValues), len(dictInst))
for cValue, listInstSeparate in dictInst.iteritems():
for inst in listInstSeparate:
self.assertEqual(cValue, inst.listAttrs[ixAttr])
def test_compute_entropy_of_split(self):
cAttrs = random.randint(2, 20)
cValues = random.randint(1, 30)
fxnGenOne = lambda _: build_entropy_one_instances(cAttrs, cValues)
fxnGenOne.cAttrs = cAttrs
fxnGenOne.cValues = cValues
fxnGenZero = build_instance_generator(0.0, cAttrs=3)
dblDelta = 0.01
for fxnGen, dblP in zip((fxnGenOne, fxnGenZero, ), (1.0, 0.0)):
listInst = fxnGen(self.cInsts)
for ixAttr in xrange(fxnGen.cAttrs):
dictInst = dtree.separate_by_attribute(listInst, ixAttr)
dblEntropy = dtree.compute_entropy_of_split(dictInst)
self.assertTrue(abs(dblEntropy - dblP) < dblDelta,
"%.3f not within %.3f of expected %.3f" %
(dblEntropy, dblDelta, dblP))
def test_compute_entropy_of_split(self):
cAttrs = random.randint(2,20)
cValues = random.randint(1,30)
fxnGenOne = lambda _: build_entropy_one_instances(cAttrs, cValues)
fxnGenOne.cAttrs = cAttrs
fxnGenOne.cValues = cValues
fxnGenZero = build_instance_generator(0.0, cAttrs=3)
dblDelta = 0.01
for fxnGen,dblP in zip((fxnGenOne,fxnGenZero,),(1.0,0.0)):
listInst = fxnGen(self.cInsts)
for ixAttr in xrange(fxnGen.cAttrs):
dictInst = dtree.separate_by_attribute(listInst, ixAttr)
dblEntropy = dtree.compute_entropy_of_split(dictInst)
self.assertTrue(abs(dblEntropy - dblP) < dblDelta,
"%.3f not within %.3f of expected %.3f" %
(dblEntropy, dblDelta, dblP))
