def test_classify(self):
"""ada_boost - classify build classifications from stumps"""
classifiers, estimates =\
ada_boost.train_dataset(self.larger_matrix,
self.larger_class_labels,
9)
data_to_classify = [1, 0.5]
classifications = ada_boost.classify(data_to_classify, classifiers)
expected = np.mat([-1.])
self.assertEqual(classifications, expected)
def test_classify(self):
"""ada_boost - classify build classifications from stumps"""
classifiers, estimates =\
ada_boost.train_dataset(self.larger_matrix,
self.larger_class_labels,
9)
data_to_classify = [1, 0.5]
classifications = ada_boost.classify(data_to_classify, classifiers)
expected = np.mat([-1.])
self.assertEqual(classifications, expected)
def test_classification():
D = np.mat(np.ones((5, 1)) / 5)
data_matrix, class_labels = load_simple_data()
stump, min_error, best_estimate = ada_boost.build_stump(data_matrix,
class_labels,
D)
logging.info('stump: {}'.format(stump))
logging.info('min_error: {}'.format(min_error))
logging.info('best_estimate: {}'.format(best_estimate))
classifier_array, aggregated_class_estimates =\
ada_boost.train_dataset(data_matrix, class_labels, 9)
data_to_classify = [0, 0]
classifications = ada_boost.classify(data_to_classify, classifier_array)
logging.info("classifications: {c}".format(c=classifications))
def test_horse_colic():
data, labels = utils.load_tsv_datafile('data/horseColicTraining.txt')
classifier_array, aggregated_class_estimates =\
ada_boost.train_dataset(data, labels, 10)
test_data, test_labels = utils.load_tsv_datafile('data/horseColicTest.txt')
prediction10 = ada_boost.classify(test_data, classifier_array)
# calculate the error
elements = 67
err_array = np.mat(np.ones((elements, 1)))
error_count = err_array[prediction10 != np.mat(test_labels).T].sum()
message = "total_errors = {total_errors}; error rate= {rate}".\
format(total_errors=error_count, rate=(error_count / elements))
logging.info(message)
def test_train_dataset(self):
"""ada_boost - train_dataset trains the dataset
and returns estimates"""
classifiers, estimates =\
ada_boost.train_dataset(self.larger_matrix,
self.larger_class_labels,
9)
expected = [
{'alpha': 0.6931471805599453,
'dim': 0,
'inequal': 'lt',
'threshold': 1.3},
{'alpha': 0.9729550745276565,
'dim': 1,
'inequal': 'lt',
'threshold': 1.0},
{'alpha': 0.8958797346140273,
'dim': 0,
'inequal': 'lt',
'threshold': 0.90000000000000002}
]
self.assertEqual(classifiers, expected)
def test_train_dataset(self):
"""ada_boost - train_dataset trains the dataset
and returns estimates"""
classifiers, estimates =\
ada_boost.train_dataset(self.larger_matrix,
self.larger_class_labels,
9)
expected = [{
'alpha': 0.6931471805599453,
'dim': 0,
'inequal': 'lt',
'threshold': 1.3
}, {
'alpha': 0.9729550745276565,
'dim': 1,
'inequal': 'lt',
'threshold': 1.0
}, {
'alpha': 0.8958797346140273,
'dim': 0,
'inequal': 'lt',
'threshold': 0.90000000000000002
}]
self.assertEqual(classifiers, expected)
def test_compute_horse_auc():
"""plot the AUC for the horse colic data"""
data, labels = utils.load_tsv_datafile('data/horseColicTraining.txt')
classifier_array, aggregated_class_estimates =\
ada_boost.train_dataset(data, labels, 40)
ada_boost.plot_roc(aggregated_class_estimates.T, labels)