def test_Compare(self):
mxKFoldRun = MxKFoldRun(5, 2)
combined5x2t = Combined5x2t()
experimentPerformance1 = mxKFoldRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.iris))
experimentPerformance2 = mxKFoldRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.iris))
self.assertAlmostEqual(
0.186,
combined5x2t.compare(experimentPerformance1,
experimentPerformance2).getPValue(), 3)
experimentPerformance1 = mxKFoldRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.tictactoe))
experimentPerformance2 = mxKFoldRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.tictactoe))
self.assertAlmostEqual(
0.0000059,
combined5x2t.compare(experimentPerformance1,
experimentPerformance2).getPValue(), 7)
experimentPerformance1 = mxKFoldRun.execute(
Experiment(Lda(), Parameter(1), self.dermatology))
experimentPerformance2 = mxKFoldRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.dermatology))
self.assertAlmostEqual(
0.9819,
combined5x2t.compare(experimentPerformance1,
experimentPerformance2).getPValue(), 4)
experimentPerformance1 = mxKFoldRun.execute(
Experiment(Dummy(), Parameter(1), self.nursery))
experimentPerformance2 = mxKFoldRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.nursery))
self.assertAlmostEqual(
0.0,
combined5x2t.compare(experimentPerformance1,
experimentPerformance2).getPValue(), 4)
experimentPerformance1 = mxKFoldRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.car))
experimentPerformance2 = mxKFoldRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.car))
self.assertAlmostEqual(
0.00043,
combined5x2t.compare(experimentPerformance1,
experimentPerformance2).getPValue(), 5)
experimentPerformance1 = mxKFoldRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.bupa))
experimentPerformance2 = mxKFoldRun.execute(
Experiment(Lda(), Parameter(1), self.bupa))
self.assertAlmostEqual(
0.0663,
combined5x2t.compare(experimentPerformance1,
experimentPerformance2).getPValue(), 4)
def test_Compare(self):
kFoldRun = KFoldRun(10)
pairedt = Pairedt()
experimentPerformance1 = kFoldRun.execute(Experiment(C45(), C45Parameter(1, True, 0.2), self.iris))
experimentPerformance2 = kFoldRun.execute(Experiment(LinearPerceptron(), LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100), self.iris))
self.assertAlmostEqual(0.379, pairedt.compare(experimentPerformance1, experimentPerformance2).getPValue(), 3)
experimentPerformance1 = kFoldRun.execute(Experiment(C45(), C45Parameter(1, True, 0.2), self.tictactoe))
experimentPerformance2 = kFoldRun.execute(Experiment(Bagging(), BaggingParameter(1, 50), self.tictactoe))
self.assertAlmostEqual(0.00000692, pairedt.compare(experimentPerformance1, experimentPerformance2).getPValue(), 7)
experimentPerformance1 = kFoldRun.execute(Experiment(Lda(), Parameter(1), self.dermatology))
experimentPerformance2 = kFoldRun.execute(Experiment(LinearPerceptron(), LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100), self.dermatology))
self.assertAlmostEqual(0.7842, pairedt.compare(experimentPerformance1, experimentPerformance2).getPValue(), 4)
experimentPerformance1 = kFoldRun.execute(Experiment(Dummy(), Parameter(1), self.nursery))
experimentPerformance2 = kFoldRun.execute(Experiment(NaiveBayes(), Parameter(1), self.nursery))
self.assertAlmostEqual(0.0, pairedt.compare(experimentPerformance1, experimentPerformance2).getPValue(), 4)
experimentPerformance1 = kFoldRun.execute(Experiment(NaiveBayes(), Parameter(1), self.car))
experimentPerformance2 = kFoldRun.execute(Experiment(Bagging(), BaggingParameter(1, 50), self.car))
self.assertAlmostEqual(0.00000336, pairedt.compare(experimentPerformance1, experimentPerformance2).getPValue(), 7)
experimentPerformance1 = kFoldRun.execute(Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()), self.bupa))
experimentPerformance2 = kFoldRun.execute(Experiment(Lda(), Parameter(1), self.bupa))
self.assertAlmostEqual(0.1640, pairedt.compare(experimentPerformance1, experimentPerformance2).getPValue(), 4)
def test_Train(self):
bagging = Bagging()
baggingParameter = BaggingParameter(1, 100)
bagging.train(self.iris.getInstanceList(), baggingParameter)
self.assertAlmostEqual(2.0, 100 * bagging.test(self.iris.getInstanceList()).getErrorRate(), 2)
bagging.train(self.bupa.getInstanceList(), baggingParameter)
self.assertAlmostEqual(42.03, 100 * bagging.test(self.bupa.getInstanceList()).getErrorRate(), 2)
bagging.train(self.dermatology.getInstanceList(), baggingParameter)
self.assertAlmostEqual(2.46, 100 * bagging.test(self.dermatology.getInstanceList()).getErrorRate(), 2)
bagging.train(self.car.getInstanceList(), baggingParameter)
self.assertAlmostEqual(0.0, 100 * bagging.test(self.car.getInstanceList()).getErrorRate(), 2)
bagging.train(self.tictactoe.getInstanceList(), baggingParameter)
self.assertAlmostEqual(0.0, 100 * bagging.test(self.tictactoe.getInstanceList()).getErrorRate(), 2)
def train(self, trainSet: InstanceList, parameters: BaggingParameter):
"""
Bagging bootstrap ensemble method that creates individuals for its ensemble by training each classifier on a
random redistribution of the training set.
This training method is for a bagged decision tree classifier. 20 percent of the instances are left aside for
pruning of the trees 80 percent of the instances are used for training the trees. The number of trees
(forestSize) is a parameter, and basically the method will learn an ensemble of trees as a model.
PARAMETERS
----------
trainSet : InstanceList
Training data given to the algorithm.
parameters : Parameter
Parameters of the bagging trees algorithm. ensembleSize returns the number of trees in the bagged forest.
"""
forestSize = parameters.getEnsembleSize()
forest = []
for i in range(forestSize):
bootstrap = trainSet.bootstrap(i)
tree = DecisionTree(
DecisionNode(InstanceList(bootstrap.getSample())))
forest.append(tree)
self.model = TreeEnsembleModel(forest)
def test_Execute(self):
mxKFoldRun = MxKFoldRun(5, 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.iris))
self.assertAlmostEqual(
6.13, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.tictactoe))
self.assertAlmostEqual(
23.51,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.bupa))
self.assertAlmostEqual(
37.05,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.dermatology))
self.assertAlmostEqual(
15.41,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Lda(), Parameter(1), self.bupa))
self.assertAlmostEqual(
34.72,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Lda(), Parameter(1), self.dermatology))
self.assertAlmostEqual(
4.04, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.iris))
self.assertAlmostEqual(
5.2, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.dermatology))
self.assertAlmostEqual(
5.46, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.car))
self.assertAlmostEqual(
16.52,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.nursery))
self.assertAlmostEqual(
9.80, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.tictactoe))
self.assertAlmostEqual(
8.77, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.car))
self.assertAlmostEqual(
9.77, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Dummy(), Parameter(1), self.nursery))
self.assertAlmostEqual(
67.09,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = mxKFoldRun.execute(
Experiment(Dummy(), Parameter(1), self.iris))
self.assertAlmostEqual(
70.53,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
def test_Execute(self):
kFoldRun = KFoldRun(10)
experimentPerformance = kFoldRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.iris))
self.assertAlmostEqual(
6.00, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.tictactoe))
self.assertAlmostEqual(
18.78,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = kFoldRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.bupa))
self.assertAlmostEqual(
36.85,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = kFoldRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.dermatology))
self.assertAlmostEqual(
10.92,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = kFoldRun.execute(
Experiment(Lda(), Parameter(1), self.bupa))
self.assertAlmostEqual(
31.61,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = kFoldRun.execute(
Experiment(Lda(), Parameter(1), self.dermatology))
self.assertAlmostEqual(
3.30, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.iris))
self.assertAlmostEqual(
5.33, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.dermatology))
self.assertAlmostEqual(
3.81, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.car))
self.assertAlmostEqual(
14.88,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = kFoldRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.nursery))
self.assertAlmostEqual(
9.71, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.tictactoe))
self.assertAlmostEqual(
3.55, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.car))
self.assertAlmostEqual(
6.77, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = kFoldRun.execute(
Experiment(Dummy(), Parameter(1), self.nursery))
self.assertAlmostEqual(
67.12,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = kFoldRun.execute(
Experiment(Dummy(), Parameter(1), self.iris))
self.assertAlmostEqual(
79.33,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
def test_Execute(self):
stratifiedMxKRun = StratifiedMxKFoldRun(5, 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.iris))
self.assertAlmostEqual(
8.00, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.tictactoe))
self.assertAlmostEqual(
22.03,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.bupa))
self.assertAlmostEqual(
33.33,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.dermatology))
self.assertAlmostEqual(
13.66,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Lda(), Parameter(1), self.bupa))
self.assertAlmostEqual(
33.05,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Lda(), Parameter(1), self.dermatology))
self.assertAlmostEqual(
3.55, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.iris))
self.assertAlmostEqual(
6.00, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.dermatology))
self.assertAlmostEqual(
4.66, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.car))
self.assertAlmostEqual(
15.11,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.nursery))
self.assertAlmostEqual(
9.68, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.tictactoe))
self.assertAlmostEqual(
9.29, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.car))
self.assertAlmostEqual(
9.20, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Dummy(), Parameter(1), self.nursery))
self.assertAlmostEqual(
66.67,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = stratifiedMxKRun.execute(
Experiment(Dummy(), Parameter(1), self.iris))
self.assertAlmostEqual(
66.67,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
def test_Execute(self):
bootstrapRun = BootstrapRun(10)
experimentPerformance = bootstrapRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.iris))
self.assertAlmostEqual(
4.33, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(C45(), C45Parameter(1, True, 0.2), self.tictactoe))
self.assertAlmostEqual(
13.16,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = bootstrapRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.bupa))
self.assertAlmostEqual(
24.84,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = bootstrapRun.execute(
Experiment(Knn(), KnnParameter(1, 3, EuclidianDistance()),
self.dermatology))
self.assertAlmostEqual(
8.01, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(Lda(), Parameter(1), self.bupa))
self.assertAlmostEqual(
32.03,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = bootstrapRun.execute(
Experiment(Lda(), Parameter(1), self.dermatology))
self.assertAlmostEqual(
2.95, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.iris))
self.assertAlmostEqual(
3.27, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(LinearPerceptron(),
LinearPerceptronParameter(1, 0.1, 0.99, 0.2, 100),
self.dermatology))
self.assertAlmostEqual(
2.65, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.car))
self.assertAlmostEqual(
14.75,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = bootstrapRun.execute(
Experiment(NaiveBayes(), Parameter(1), self.nursery))
self.assertAlmostEqual(
9.71, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.tictactoe))
self.assertAlmostEqual(
3.00, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(Bagging(), BaggingParameter(1, 50), self.car))
self.assertAlmostEqual(
3.44, 100 * experimentPerformance.meanPerformance().getErrorRate(),
2)
experimentPerformance = bootstrapRun.execute(
Experiment(Dummy(), Parameter(1), self.nursery))
self.assertAlmostEqual(
66.79,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)
experimentPerformance = bootstrapRun.execute(
Experiment(Dummy(), Parameter(1), self.iris))
self.assertAlmostEqual(
66.67,
100 * experimentPerformance.meanPerformance().getErrorRate(), 2)