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
|
from functools import partial
import pytest
from sklearn.tree import DecisionTreeClassifier
from skopt import forest_minimize, gbrt_minimize
from skopt.benchmarks import bench1, bench2, bench3, bench4
MINIMIZERS = [
("ET", partial(forest_minimize, base_estimator='ET')),
("RF", partial(forest_minimize, base_estimator='RF')),
("gbrt", gbrt_minimize),
]
@pytest.mark.fast_test
@pytest.mark.parametrize("base_estimator", [42, DecisionTreeClassifier()])
def test_forest_minimize_api(base_estimator):
# invalid string value
with pytest.raises(ValueError):
forest_minimize(lambda x: 0.0, [], base_estimator='abc')
# not a string nor a regressor
with pytest.raises(ValueError):
forest_minimize(lambda x: 0.0, [], base_estimator=base_estimator)
def check_minimize(
minimizer,
func,
y_opt,
dimensions,
margin,
n_calls,
n_initial_points=10,
x0=None,
n_jobs=1,
):
for n in range(3):
r = minimizer(
func,
dimensions,
n_calls=n_calls,
random_state=n,
n_initial_points=n_initial_points,
x0=x0,
n_jobs=n_jobs,
)
assert r.fun < y_opt + margin
@pytest.mark.slow_test
@pytest.mark.parametrize("name, minimizer", MINIMIZERS)
def test_tree_based_minimize(name, minimizer):
check_minimize(minimizer, bench1, 0.05, [(-2.0, 2.0)], 0.05, 25, 5)
# XXX: We supply points at the edge of the search
# space as an initial point to the minimizer.
# This makes sure that the RF model can find the minimum even
# if all the randomly sampled points are one side of the
# the minimum, since for a decision tree any point greater than
# max(sampled_points) would give a constant value.
X0 = [[-5.6], [-5.8], [5.8], [5.6]]
check_minimize(minimizer, bench2, -4.7, [(-6.0, 6.0)], 0.1, 20, 10, X0)
check_minimize(minimizer, bench3, -0.4, [(-2.0, 2.0)], 0.05, 10, 5)
check_minimize(minimizer, bench4, 1.0, [["-2", "-1", "0", "1", "2"]], 0.05, 5, 1)
@pytest.mark.slow_test
def test_tree_based_minimize_n_jobs():
check_minimize(forest_minimize, bench1, 0.05, [(-2.0, 2.0)], 0.05, 25, 5, n_jobs=2)
@pytest.mark.fast_test
def test_categorical_integer():
def f(params):
return 0
dims = [[1]]
res = forest_minimize(f, dims, n_calls=1, random_state=1, n_initial_points=1)
assert res.x_iters[0][0] == dims[0][0]
|
