1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
|
from itertools import chain
from random import randint
from typing import Tuple, Dict, Callable, Type
from AnyQt.QtCore import Qt
from AnyQt.QtGui import QStandardItem, QStandardItemModel
from AnyQt.QtWidgets import QWidget, QVBoxLayout
from Orange.base import Learner
from Orange.data import Table
from Orange.modelling import GBLearner
try:
from Orange.modelling import CatGBLearner
except ImportError:
CatGBLearner = None
try:
from Orange.modelling import XGBLearner, XGBRFLearner
except ImportError:
XGBLearner = XGBRFLearner = None
from Orange.widgets import gui
from Orange.widgets.settings import Setting, SettingProvider
from Orange.widgets.utils.owlearnerwidget import OWBaseLearner
from Orange.widgets.utils.widgetpreview import WidgetPreview
class LearnerItemModel(QStandardItemModel):
LEARNERS = [
(GBLearner, "", ""),
(XGBLearner, "Extreme Gradient Boosting (xgboost)", "xgboost"),
(XGBRFLearner, "Extreme Gradient Boosting Random Forest (xgboost)",
"xgboost"),
(CatGBLearner, "Gradient Boosting (catboost)", "catboost"),
]
def __init__(self, parent):
super().__init__(parent)
self._add_data()
def _add_data(self):
for cls, opt_name, lib in self.LEARNERS:
item = QStandardItem()
imported = bool(cls)
name = cls.name if imported else opt_name
item.setData(f"{name}", Qt.DisplayRole)
item.setEnabled(imported)
if not imported:
item.setToolTip(f"{lib} is not installed")
self.appendRow(item)
class BaseEditor(QWidget, gui.OWComponent):
learner_class: Type[Learner] = NotImplemented
n_estimators: int = NotImplemented
learning_rate: float = NotImplemented
random_state: bool = NotImplemented
max_depth: int = NotImplemented
def __init__(self, parent: OWBaseLearner):
QWidget.__init__(self, parent)
gui.OWComponent.__init__(self, parent)
self.settings_changed: Callable = parent.settings_changed
self.setLayout(QVBoxLayout())
self.layout().setContentsMargins(0, 0, 0, 0)
self._layout: QWidget = gui.vBox(self, spacing=6, margin=0)
self._add_main_layout()
def _add_main_layout(self):
common_args = {"callback": self.settings_changed,
"alignment": Qt.AlignRight, "controlWidth": 80}
self.basic_box = gui.vBox(self._layout, "Basic Properties")
gui.spin(
self.basic_box, self, "n_estimators", 1, 10000,
label="Number of trees:", **common_args
)
gui.doubleSpin(
self.basic_box, self, "learning_rate", 0, 1, 0.001,
label="Learning rate: ", **common_args
)
gui.checkBox(
self.basic_box, self, "random_state", label="Replicable training",
callback=self.settings_changed, attribute=Qt.WA_LayoutUsesWidgetRect
)
self.growth_box = gui.vBox(self._layout, "Growth Control")
gui.spin(
self.growth_box, self, "max_depth", 1, 50,
label="Limit depth of individual trees: ", **common_args
)
self.sub_box = gui.vBox(self._layout, "Subsampling")
def get_arguments(self) -> Dict:
return {
"n_estimators": self.n_estimators,
"learning_rate": self.learning_rate,
"random_state": 0 if self.random_state else randint(1, 1000000),
"max_depth": self.max_depth,
}
def get_learner_parameters(self) -> Tuple:
return (
("Method", self.learner_class.name),
("Number of trees", self.n_estimators),
("Learning rate", self.learning_rate),
("Replicable training", "Yes" if self.random_state else "No"),
("Maximum tree depth", self.max_depth),
)
class RegEditor(BaseEditor):
LAMBDAS = list(chain([x / 10000 for x in range(1, 10)],
[x / 1000 for x in range(1, 20)],
[x / 100 for x in range(2, 20)],
[x / 10 for x in range(2, 9)],
range(1, 20),
range(20, 100, 5),
range(100, 1001, 100)))
lambda_index: int = NotImplemented
@property
def lambda_(self):
return self.LAMBDAS[int(self.lambda_index)]
def _add_main_layout(self):
super()._add_main_layout()
# Basic properties
box = self.basic_box
gui.separator(box, height=1)
gui.widgetLabel(box, "Regularization:")
gui.hSlider(
box, self, "lambda_index", minValue=0, createLabel=False,
maxValue=len(self.LAMBDAS) - 1, callback=self._set_lambda_label,
callback_finished=self.settings_changed
)
box2 = gui.hBox(box)
box2.layout().setAlignment(Qt.AlignCenter)
self.lambda_label = gui.widgetLabel(box2, "")
self._set_lambda_label()
def _set_lambda_label(self):
self.lambda_label.setText("Lambda: {}".format(self.lambda_))
def get_arguments(self) -> Dict:
params = super().get_arguments()
params["reg_lambda"] = self.lambda_
return params
def get_learner_parameters(self) -> Tuple:
return super().get_learner_parameters() + (
("Regularization strength", self.lambda_),
)
class GBLearnerEditor(BaseEditor):
learner_class = GBLearner
n_estimators = Setting(100)
learning_rate = Setting(0.1)
random_state = Setting(True)
subsample = Setting(1)
max_depth = Setting(3)
min_samples_split = Setting(2)
def _add_main_layout(self):
super()._add_main_layout()
# Subsampling
gui.doubleSpin(
self.sub_box, self, "subsample", 0.05, 1, 0.05,
controlWidth=80, alignment=Qt.AlignRight,
label="Fraction of training instances: ",
callback=self.settings_changed
)
# Growth control
gui.spin(
self.growth_box, self, "min_samples_split", 2, 1000,
controlWidth=80, label="Do not split subsets smaller than: ",
alignment=Qt.AlignRight, callback=self.settings_changed
)
def get_arguments(self) -> Dict:
params = super().get_arguments()
params["subsample"] = self.subsample
params["min_samples_split"] = self.min_samples_split
return params
def get_learner_parameters(self) -> Tuple:
return super().get_learner_parameters() + (
("Fraction of training instances", self.subsample),
("Stop splitting nodes with maximum instances",
self.min_samples_split),
)
class CatGBLearnerEditor(RegEditor):
learner_class = CatGBLearner
n_estimators = Setting(100)
learning_rate = Setting(0.3)
random_state = Setting(True)
max_depth = Setting(6)
lambda_index = Setting(55) # 3
colsample_bylevel = Setting(1)
def _add_main_layout(self):
super()._add_main_layout()
# Subsampling
gui.doubleSpin(
self.sub_box, self, "colsample_bylevel", 0.05, 1, 0.05,
controlWidth=80, alignment=Qt.AlignRight,
label="Fraction of features for each tree: ",
callback=self.settings_changed
)
def get_arguments(self) -> Dict:
params = super().get_arguments()
params["colsample_bylevel"] = self.colsample_bylevel
return params
def get_learner_parameters(self) -> Tuple:
return super().get_learner_parameters() + (
("Fraction of features for each tree", self.colsample_bylevel),
)
class XGBBaseEditor(RegEditor):
learner_class = XGBLearner
n_estimators = Setting(100)
learning_rate = Setting(0.3)
random_state = Setting(True)
max_depth = Setting(6)
lambda_index = Setting(53) # 1
subsample = Setting(1)
colsample_bytree = Setting(1)
colsample_bylevel = Setting(1)
colsample_bynode = Setting(1)
def _add_main_layout(self):
super()._add_main_layout()
# Subsampling
common_args = {"callback": self.settings_changed,
"alignment": Qt.AlignRight, "controlWidth": 80}
gui.doubleSpin(
self.sub_box, self, "subsample", 0.05, 1, 0.05,
label="Fraction of training instances: ", **common_args
)
gui.doubleSpin(
self.sub_box, self, "colsample_bytree", 0.05, 1, 0.05,
label="Fraction of features for each tree: ", **common_args
)
gui.doubleSpin(
self.sub_box, self, "colsample_bylevel", 0.05, 1, 0.05,
label="Fraction of features for each level: ", **common_args
)
gui.doubleSpin(
self.sub_box, self, "colsample_bynode", 0.05, 1, 0.05,
label="Fraction of features for each split: ", **common_args
)
def get_arguments(self) -> Dict:
params = super().get_arguments()
params["subsample"] = self.subsample
params["colsample_bytree"] = self.colsample_bytree
params["colsample_bylevel"] = self.colsample_bylevel
params["colsample_bynode"] = self.colsample_bynode
return params
def get_learner_parameters(self) -> Tuple:
return super().get_learner_parameters() + (
("Fraction of training instances", self.subsample),
("Fraction of features for each tree", self.colsample_bytree),
("Fraction of features for each level", self.colsample_bylevel),
("Fraction of features for each split", self.colsample_bynode),
)
class XGBLearnerEditor(XGBBaseEditor):
learner_class = XGBLearner
class XGBRFLearnerEditor(XGBBaseEditor):
learner_class = XGBRFLearner
class OWGradientBoosting(OWBaseLearner):
name = "Gradient Boosting"
description = "Predict using gradient boosting on decision trees."
icon = "icons/GradientBoosting.svg"
priority = 45
keywords = "gradient boosting, catboost, gradient, boost, tree, forest, xgb, gb, extreme"
LEARNER: Learner = GBLearner
editor: BaseEditor = None
gb_editor = SettingProvider(GBLearnerEditor)
xgb_editor = SettingProvider(XGBLearnerEditor)
xgbrf_editor = SettingProvider(XGBRFLearnerEditor)
catgb_editor = SettingProvider(CatGBLearnerEditor)
method_index = Setting(0)
def add_main_layout(self):
# this is part of init, pylint: disable=attribute-defined-outside-init
box = gui.vBox(self.controlArea, "Method")
gui.comboBox(
box, self, "method_index", model=LearnerItemModel(self),
callback=self.__method_changed
)
self.gb_editor = GBLearnerEditor(self)
self.xgb_editor = XGBLearnerEditor(self)
self.xgbrf_editor = XGBRFLearnerEditor(self)
self.catgb_editor = CatGBLearnerEditor(self)
self.editors = [self.gb_editor, self.xgb_editor,
self.xgbrf_editor, self.catgb_editor]
editor_box = gui.widgetBox(self.controlArea)
for editor in self.editors:
editor_box.layout().addWidget(editor)
editor.hide()
if self.editors[int(self.method_index)].learner_class is None:
self.method_index = 0
self.editor = self.editors[int(self.method_index)]
self.editor.show()
def __method_changed(self):
self.editor.hide()
self.editor = self.editors[int(self.method_index)]
self.editor.show()
self.settings_changed()
def create_learner(self) -> Learner:
learner = self.editor.learner_class
kwargs = self.editor.get_arguments()
return learner(preprocessors=self.preprocessors, **kwargs)
def get_learner_parameters(self) -> Tuple:
return self.editor.get_learner_parameters()
if __name__ == "__main__": # pragma: no cover
WidgetPreview(OWGradientBoosting).run(Table("iris"))
|
