|
1
|
|
|
# Author: Simon Blanke |
|
2
|
|
|
# Email: [email protected] |
|
3
|
|
|
# License: MIT License |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
from sklearn.base import BaseEstimator, clone |
|
7
|
|
|
from sklearn.metrics import check_scoring |
|
8
|
|
|
from sklearn.utils.validation import indexable, _check_method_params |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
from hyperactive import Hyperactive |
|
12
|
|
|
|
|
13
|
|
|
from .objective_function_adapter import ObjectiveFunctionAdapter |
|
14
|
|
|
from .best_estimator import BestEstimator |
|
15
|
|
|
|
|
16
|
|
|
|
|
17
|
|
|
class HyperactiveSearchCV(BaseEstimator, BestEstimator): |
|
18
|
|
|
_required_parameters = ["estimator", "optimizer", "params_config"] |
|
19
|
|
|
|
|
20
|
|
|
def __init__( |
|
21
|
|
|
self, |
|
22
|
|
|
estimator, |
|
23
|
|
|
optimizer, |
|
24
|
|
|
params_config, |
|
25
|
|
|
n_iter=100, |
|
26
|
|
|
*, |
|
27
|
|
|
scoring=None, |
|
28
|
|
|
n_jobs=1, |
|
29
|
|
|
random_state=None, |
|
30
|
|
|
refit=True, |
|
31
|
|
|
cv=None, |
|
32
|
|
|
): |
|
33
|
|
|
self.estimator = estimator |
|
34
|
|
|
self.optimizer = optimizer |
|
35
|
|
|
self.params_config = params_config |
|
36
|
|
|
self.n_iter = n_iter |
|
37
|
|
|
self.scoring = scoring |
|
38
|
|
|
self.n_jobs = n_jobs |
|
39
|
|
|
self.random_state = random_state |
|
40
|
|
|
self.refit = refit |
|
41
|
|
|
self.cv = cv |
|
42
|
|
|
|
|
43
|
|
|
def _refit( |
|
44
|
|
|
self, |
|
45
|
|
|
X, |
|
46
|
|
|
y=None, |
|
47
|
|
|
**fit_params, |
|
48
|
|
|
): |
|
49
|
|
|
self.best_estimator_ = clone(self.estimator) |
|
50
|
|
|
self.best_estimator_.fit(X, y, **fit_params) |
|
51
|
|
|
return self |
|
52
|
|
|
|
|
53
|
|
|
def fit(self, X, y, **params): |
|
54
|
|
|
X, y = indexable(X, y) |
|
55
|
|
|
X, y = self._validate_data(X, y) |
|
56
|
|
|
|
|
57
|
|
|
params = _check_method_params(X, params=params) |
|
58
|
|
|
self.scorer_ = check_scoring(self.estimator, scoring=self.scoring) |
|
59
|
|
|
|
|
60
|
|
|
objective_function_adapter = ObjectiveFunctionAdapter( |
|
61
|
|
|
self.estimator, |
|
62
|
|
|
) |
|
63
|
|
|
objective_function_adapter.add_dataset(X, y) |
|
64
|
|
|
objective_function_adapter.add_validation(self.scorer_, self.cv) |
|
65
|
|
|
|
|
66
|
|
|
hyper = Hyperactive(verbosity=False) |
|
67
|
|
|
hyper.add_search( |
|
68
|
|
|
objective_function_adapter.objective_function, |
|
69
|
|
|
search_space=self.params_config, |
|
70
|
|
|
optimizer=self.optimizer, |
|
71
|
|
|
n_iter=self.n_iter, |
|
72
|
|
|
n_jobs=self.n_jobs, |
|
73
|
|
|
random_state=self.random_state, |
|
74
|
|
|
) |
|
75
|
|
|
hyper.run() |
|
76
|
|
|
|
|
77
|
|
|
if self.refit: |
|
78
|
|
|
self._refit(X, y, **params) |
|
79
|
|
|
|
|
80
|
|
|
return self |
|
81
|
|
|
|
|
82
|
|
|
def score(self, X, y=None, **params): |
|
83
|
|
|
return self.scorer_(self.best_estimator_, X, y, **params) |
|
84
|
|
|
|
SimonBlanke /
Hyperactive
