def test1(self):
# " testing pruning with known results "
oPts = [
[0, 0, 1, 0],
[0, 1, 1, 1],
[1, 0, 1, 1],
[1, 1, 0, 0],
[1, 1, 1, 1],
]
tPts = oPts + [[0, 1, 1, 0], [0, 1, 1, 0]]
tree = ID3.ID3Boot(oPts, attrs=range(3), nPossibleVals=[2] * 4)
err, badEx = CrossValidate.CrossValidate(tree, oPts)
assert err == 0.0, 'bad initial error'
assert len(badEx) == 0, 'bad initial error'
# prune with original data, shouldn't do anything
f = StringIO()
with redirect_stdout(f):
PruneTree._verbose = True
newTree, err = PruneTree.PruneTree(tree, [], oPts)
PruneTree._verbose = False
self.assertIn('Pruner', f.getvalue())
assert newTree == tree, 'improper pruning'
# prune with train data
newTree, err = PruneTree.PruneTree(tree, [], tPts)
assert newTree != tree, 'bad pruning'
assert feq(err, 0.14286), 'bad error result'
def _testChain():
from rdkit.ML.DecTree import ID3
oPts= [ \
[1,0,0,0,1],
[1,0,0,0,1],
[1,0,0,0,1],
[1,0,0,0,1],
[1,0,0,0,1],
[1,0,0,0,1],
[1,0,0,0,1],
[0,0,1,1,0],
[0,0,1,1,0],
[0,0,1,1,1],
[0,1,0,1,0],
[0,1,0,1,0],
[0,1,0,0,1],
]
tPts = oPts
tree = ID3.ID3Boot(oPts,
attrs=range(len(oPts[0]) - 1),
nPossibleVals=[2] * len(oPts[0]))
tree.Print()
err, badEx = CrossValidate.CrossValidate(tree, oPts)
print('original error:', err)
err, badEx = CrossValidate.CrossValidate(tree, tPts)
print('original holdout error:', err)
newTree, frac2 = PruneTree(tree, oPts, tPts)
newTree.Print()
err, badEx = CrossValidate.CrossValidate(newTree, tPts)
print('pruned holdout error is:', err)
print(badEx)
def _testSpecific():
from rdkit.ML.DecTree import ID3
oPts= [ \
[0,0,1,0],
[0,1,1,1],
[1,0,1,1],
[1,1,0,0],
[1,1,1,1],
]
tPts = oPts + [[0, 1, 1, 0], [0, 1, 1, 0]]
tree = ID3.ID3Boot(oPts, attrs=range(3), nPossibleVals=[2] * 4)
tree.Print()
err, badEx = CrossValidate.CrossValidate(tree, oPts)
print('original error:', err)
err, badEx = CrossValidate.CrossValidate(tree, tPts)
print('original holdout error:', err)
newTree, frac2 = PruneTree(tree, oPts, tPts)
newTree.Print()
err, badEx = CrossValidate.CrossValidate(newTree, tPts)
print('pruned holdout error is:', err)
print(badEx)
print(len(tree), len(newTree))
def _setupMultiTree(self):
examples = [[0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 2], [0, 1, 1, 2],
[1, 0, 0, 2], [1, 0, 1, 2], [1, 1, 0, 2], [1, 1, 1, 0]]
data = MLData.MLQuantDataSet(examples)
attrs = range(0, data.GetNVars())
t1 = ID3.ID3Boot(data.GetAllData(), attrs, data.GetNPossibleVals())
self.t1 = t1
self.examples = examples
def TestTree():
""" testing code for named trees
"""
examples1 = [['p1', 0, 1, 0, 0], ['p2', 0, 0, 0, 1], ['p3', 0, 0, 1, 2], ['p4', 0, 1, 1, 2],
['p5', 1, 0, 0, 2], ['p6', 1, 0, 1, 2], ['p7', 1, 1, 0, 2], ['p8', 1, 1, 1, 0]]
attrs = list(range(1, len(examples1[0]) - 1))
nPossibleVals = [0, 2, 2, 2, 3]
t1 = ID3.ID3Boot(examples1, attrs, nPossibleVals, maxDepth=1)
t1.Print()
def _setupPyMultiTree(self):
from rdkit.ML.InfoTheory import entropy
ID3.entropy.InfoEntropy = entropy.PyInfoEntropy
ID3.entropy.InfoGain = entropy.PyInfoGain
examples = [[0, 1, 0, 0], [0, 0, 0, 1], [0, 0, 1, 2], [0, 1, 1, 2],
[1, 0, 0, 2], [1, 0, 1, 2], [1, 1, 0, 2], [1, 1, 1, 0]]
data = MLData.MLQuantDataSet(examples)
attrs = range(0, data.GetNVars())
t1 = ID3.ID3Boot(data.GetAllData(), attrs, data.GetNPossibleVals())
self.t1 = t1
self.examples = examples
def test1(self):
" testing pruning with known results "
oPts= [ \
[0,0,1,0],
[0,1,1,1],
[1,0,1,1],
[1,1,0,0],
[1,1,1,1],
]
tPts = oPts+[[0,1,1,0],[0,1,1,0]]
tree = ID3.ID3Boot(oPts,attrs=range(3),nPossibleVals=[2]*4)
err,badEx = CrossValidate.CrossValidate(tree,oPts)
assert err==0.0,'bad initial error'
assert len(badEx)==0,'bad initial error'
# prune with original data, shouldn't do anything
newTree,err = PruneTree.PruneTree(tree,[],oPts)
assert newTree==tree,'improper pruning'
# prune with train data
newTree,err = PruneTree.PruneTree(tree,[],tPts)
assert newTree!=tree,'bad pruning'
assert feq(err,0.14286),'bad error result'
def GenRandomExamples(nVars=10, randScale=0.3, bitProb=0.5, nExamples=500, seed=(0, 0),
addResults=1):
random.seed(seed[0])
varWeights = numpy.array([random.random() for _ in range(nVars)]) * randScale
examples = [None] * nExamples
for i in range(nExamples):
varVals = [random.random() > bitProb for _ in range(nVars)]
temp = numpy.array(varVals) * varWeights
res = sum(temp)
if addResults:
varVals.append(res >= 1.)
examples[i] = varVals
nPossibleVals = [2] * (nExamples + 1)
attrs = list(range(nVars))
return (examples, attrs, nPossibleVals)
if __name__ == '__main__': # pragma: nocover
from rdkit.six.moves import cPickle
examples, attrs, nPossibleVals = GenRandomExamples()
outF = open('random.dat.pkl', 'wb+')
cPickle.dump(examples, outF)
cPickle.dump(attrs, outF)
cPickle.dump(nPossibleVals, outF)
tree = ID3.ID3Boot(examples, attrs, nPossibleVals)
tree.Pickle('save.pkl')