def test_ensemble(self):
X, y = make_blob_data()
clf = AllStumpsClassifier(X, y)
self.assertGreaterEqual(clf.score(X, y), 0)
def test_decision_stump(self):
X, y = make_blob_data()
stump = DecisionStumpClassifier(X, y, 2)
# Basic sanity check:
self.assertEqual(len(stump.predict(X)), len(X))
def setUpClass(cls):
cls.X, cls.y = make_blob_data()
cls.clf = QBoostClassifier(cls.X, cls.y, 0.0)
'digits', help='handwritten digits data set')
sp_digits.add_argument('--digit1', type=int, default=0, choices=range(10),
help='first digit to include (default: %(default)s)')
sp_digits.add_argument('--digit2', type=int, default=1, choices=range(10),
help='second digit to include (default: %(default)s)')
sp_digits.add_argument('--plot-digits', action='store_true',
help='plot a random sample of each digit')
args = parser.parse_args()
if args.dataset == 'blobs':
n_samples = args.num_samples
n_features = args.num_features
n_informative = args.num_informative
X, y = make_blob_data(
n_samples=n_samples, n_features=n_features, n_informative=n_informative)
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.4)
if args.cross_validation:
# See Boyda et al. (2017), Eq. (17) regarding normalization
normalized_lambdas = np.linspace(0.0, 0.5, 10)
lambdas = normalized_lambdas / n_features
print('Performing cross-validation using {} values of lambda, this may take several minutes...'.format(len(lambdas)))
qboost, lam = qboost_lambda_sweep(
X_train, y_train, lambdas, verbose=args.verbose)
else:
qboost = QBoostClassifier(X_train, y_train, args.lam)
if args.verbose: