def test_learner_with_transformation(self):
learner = RandomForestLearner(random_state=0)
from Orange.projection import PCA
iris = Table("iris")
data = PCA(n_components=2)(iris)(iris)
scores = learner.score_data(data)
np.testing.assert_almost_equal(scores, [[0.7760495, 0.2239505]])
def test_learner_with_transformation(self):
learner = RandomForestLearner(random_state=0)
from Orange.projection import PCA
iris = Table("iris")
data = PCA(n_components=2)(iris)(iris)
scores = learner.score_data(data)
np.testing.assert_almost_equal(scores, [[0.7760495, 0.2239505]])
def test_input_scorer_fitter(self):
heart_disease = Table('heart_disease')
self.assertEqual(self.widget.scorers, {})
model = self.widget.ranksModel
for fitter, name in ((RandomForestLearner(), 'random forest'),
(SGDLearner(), 'sgd')):
with self.subTest(fitter=fitter),\
warnings.catch_warnings():
warnings.filterwarnings("ignore", ".*", ConvergenceWarning)
self.send_signal("Scorer", fitter, 1)
for data in (self.housing, heart_disease):
with self.subTest(data=data.name):
self.send_signal('Data', data)
self.wait_until_finished()
scores = [
model.data(
model.index(row,
model.columnCount() - 1))
for row in range(model.rowCount())
]
self.assertEqual(len(scores),
len(data.domain.attributes))
self.assertFalse(np.isnan(scores).any())
last_column = model.headerData(model.columnCount() - 1,
Qt.Horizontal).lower()
self.assertIn(name, last_column)
self.send_signal("Scorer", None, 1)
self.assertEqual(self.widget.scorers, {})
def test_input_scorer_fitter(self):
heart_disease = Table("heart_disease")
self.assertEqual(self.widget.scorers, {})
model = self.widget.ranksModel
for fitter, name in (
(RandomForestLearner(), "random forest"),
(SGDLearner(), "sgd"),
):
with self.subTest(fitter=fitter):
self.send_signal("Scorer", fitter, 1)
for data in (self.housing, heart_disease):
with self.subTest(data=data.name):
self.send_signal("Data", data)
scores = [
model.data(
model.index(row,
model.columnCount() - 1))
for row in range(model.rowCount())
]
self.assertEqual(len(scores),
len(data.domain.attributes))
self.assertFalse(np.isnan(scores).any())
last_column = model.headerData(model.columnCount() - 1,
Qt.Horizontal).lower()
self.assertIn(name, last_column)
self.send_signal("Scorer", None, 1)
self.assertEqual(self.widget.scorers, {})
def test_no_class_data_learner_class_reg(self):
"""
Check workflow with learners that can be both classifier
or regressor and data have no class variable. This test should not
fail.
"""
data = Table.from_table(Domain(self.iris.domain.variables), self.iris)
random_forest = RandomForestLearner()
self.assertIsNone(data.domain.class_var)
self.send_signal(self.widget.Inputs.data, data)
with patch("Orange.widgets.data.owrank.log.error") as log:
self.send_signal(self.widget.Inputs.scorer, random_forest, 1)
log.assert_called()
self.assertEqual(self.get_output(self.widget.Outputs.scores).X.shape,
(len(self.iris.domain.variables), 1))
self._trees = model.trees
def get_trees(self):
"""Get the tree adapters in the random forest."""
if not self._adapters:
self._adapters = list(map(SklTreeAdapter, self._trees))
return self._adapters
@property
def domain(self):
"""Get the domain."""
return self._domain
if __name__ == '__main__':
from Orange.modelling import RandomForestLearner
from AnyQt.QtWidgets import QApplication
import sys
app = QApplication(sys.argv)
ow = OWPythagoreanForest()
data = Table(sys.argv[1] if len(sys.argv) > 1 else 'iris')
rf = RandomForestLearner(n_estimators=10)(data)
rf.instances = data
ow.set_rf(rf)
ow.show()
ow.raise_()
ow.handleNewSignals()
app.exec_()
self._trees = model.trees
def get_trees(self):
"""Get the tree adapters in the random forest."""
if not self._adapters:
self._adapters = list(map(SklTreeAdapter, self._trees))
return self._adapters
@property
def domain(self):
"""Get the domain."""
return self._domain
if __name__ == '__main__':
from Orange.modelling import RandomForestLearner
from AnyQt.QtWidgets import QApplication
import sys
app = QApplication(sys.argv)
ow = OWPythagoreanForest()
data = Table(sys.argv[1] if len(sys.argv) > 1 else 'iris')
rf = RandomForestLearner(n_estimators=10)(data)
rf.instances = data
ow.set_rf(rf)
ow.show()
ow.raise_()
ow.handleNewSignals()
app.exec_()
class OWCatBoost(OWBaseLearner):
name = "CatBoost"
description = "An ensemble meta-algorithm"
icon = "icons/mywidget.svg"
category = "Model"
priority = 80
LEARNER = CatBoostLearner
DEFAULT_BASE_ESTIMATOR = RandomForestLearner()
random_seed = Setting(0)
n_iterations = Setting(50)
depth = Setting(6)
learning_rate = Setting(.03)
loss_index = Setting(0)
border = Setting(.5)
use_random_seed = Setting(False)
train_dir = Setting(tempfile.gettempdir())
losses = ["Logloss", "MultiClass"]
class Error(OWBaseLearner.Error):
no_weight_support = Msg('The base learner does not support weights.')
class Inputs(OWBaseLearner.Inputs):
learner = Input("Learner", Learner)
def add_main_layout(self):
box = gui.widgetBox(self.controlArea, "Parameters")
self.base_estimator = self.DEFAULT_BASE_ESTIMATOR
self.border_spin = gui.doubleSpin(box,
self,
"border",
0,
1.0,
.1,
label="Border:",
decimals=2,
alignment=Qt.AlignRight,
controlWidth=80,
callback=self.settings_changed)
self.base_label = gui.label(
box, self, "Base learner: " + self.base_estimator.name.title()
if self.base_estimator is not None else 'None')
self.path_edit = gui.lineEdit(box,
self,
"train_dir",
label="Path for training:",
alignment=Qt.AlignRight,
callback=self.settings_changed)
self.n_estimators_spin = gui.spin(box,
self,
"n_iterations",
1,
100,
label="Number of estimators:",
alignment=Qt.AlignRight,
controlWidth=80,
callback=self.settings_changed)
self.depth_spin = gui.spin(box,
self,
"depth",
1,
100,
label="Depth:",
alignment=Qt.AlignRight,
controlWidth=80,
callback=self.settings_changed,
posttext="Recommended range: [1;10]")
self.learning_rate_spin = gui.doubleSpin(
box,
self,
"learning_rate",
1e-5,
1.0,
1e-5,
label="Learning rate:",
decimals=5,
alignment=Qt.AlignRight,
controlWidth=80,
callback=self.settings_changed,
posttext="Smaller the value, the more iterations")
self.random_seed_spin = gui.spin(
box,
self,
"random_seed",
0,
2**31 - 1,
controlWidth=80,
label="Fixed seed for random generator:",
alignment=Qt.AlignRight,
callback=self.settings_changed,
checked="use_random_seed",
checkCallback=self.settings_changed)
def create_learner(self):
if self.base_estimator is None:
return None
return self.LEARNER(depth=self.depth,
base_estimator=self.base_estimator,
iterations=self.n_iterations,
learning_rate=self.learning_rate,
random_seed=self.random_seed,
preprocessors=self.preprocessors,
border=self.border)
@Inputs.learner
def set_base_learner(self, learner):
if learner and not learner.supports_weights:
# Clear the error and reset to default base learner
self.Error.no_weight_support()
self.base_estimator = None
self.base_label.setText("Base estimator: INVALID")
else:
self.base_estimator = learner or self.DEFAULT_BASE_ESTIMATOR
self.base_label.setText("Base estimator: %s" %
self.base_estimator.name.title())
if self.auto_apply:
self.apply()