|
1
|
|
|
import time |
|
2
|
|
|
import numpy as np |
|
3
|
|
|
import pandas as pd |
|
4
|
|
|
from sklearn.datasets import load_breast_cancer |
|
5
|
|
|
from sklearn.model_selection import cross_val_score |
|
6
|
|
|
from sklearn.tree import DecisionTreeClassifier |
|
7
|
|
|
from gradient_free_optimizers import RandomSearchOptimizer, HillClimbingOptimizer |
|
8
|
|
|
|
|
9
|
|
|
|
|
10
|
|
|
def objective_function(para): |
|
11
|
|
|
score = -para["x1"] * para["x1"] |
|
12
|
|
|
return score |
|
13
|
|
|
|
|
14
|
|
|
|
|
15
|
|
|
search_space = { |
|
16
|
|
|
"x1": np.arange(0, 100000, 0.1), |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
err = 0.01 |
|
20
|
|
|
|
|
21
|
|
|
|
|
22
|
|
|
def test_random_state_0(): |
|
23
|
|
|
opt0 = RandomSearchOptimizer(search_space) |
|
24
|
|
|
opt0.search( |
|
25
|
|
|
objective_function, n_iter=100, initialize={"random": 1}, random_state=1 |
|
26
|
|
|
) |
|
27
|
|
|
|
|
28
|
|
|
opt1 = RandomSearchOptimizer(search_space) |
|
29
|
|
|
opt1.search( |
|
30
|
|
|
objective_function, n_iter=100, initialize={"random": 1}, random_state=1 |
|
31
|
|
|
) |
|
32
|
|
|
|
|
33
|
|
|
assert abs(opt0.best_score - opt1.best_score) < err |
|
34
|
|
|
|
|
35
|
|
|
|
|
36
|
|
|
def test_random_state_1(): |
|
37
|
|
|
opt0 = RandomSearchOptimizer(search_space) |
|
38
|
|
|
opt0.search( |
|
39
|
|
|
objective_function, n_iter=100, initialize={"random": 1}, random_state=10 |
|
40
|
|
|
) |
|
41
|
|
|
|
|
42
|
|
|
opt1 = RandomSearchOptimizer(search_space) |
|
43
|
|
|
opt1.search( |
|
44
|
|
|
objective_function, n_iter=100, initialize={"random": 1}, random_state=10 |
|
45
|
|
|
) |
|
46
|
|
|
|
|
47
|
|
|
assert abs(opt0.best_score - opt1.best_score) < err |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
def test_random_state_2(): |
|
51
|
|
|
opt0 = RandomSearchOptimizer(search_space) |
|
52
|
|
|
opt0.search( |
|
53
|
|
|
objective_function, n_iter=100, initialize={"random": 1}, random_state=1 |
|
54
|
|
|
) |
|
55
|
|
|
|
|
56
|
|
|
opt1 = RandomSearchOptimizer(search_space) |
|
57
|
|
|
opt1.search( |
|
58
|
|
|
objective_function, n_iter=100, initialize={"random": 1}, random_state=10 |
|
59
|
|
|
) |
|
60
|
|
|
|
|
61
|
|
|
assert abs(opt0.best_score - opt1.best_score) > err |
|
62
|
|
|
|
|
63
|
|
|
|
|
64
|
|
|
def test_no_random_state_0(): |
|
65
|
|
|
opt0 = RandomSearchOptimizer(search_space) |
|
66
|
|
|
opt0.search(objective_function, n_iter=100, initialize={"random": 1}) |
|
67
|
|
|
|
|
68
|
|
|
opt1 = RandomSearchOptimizer(search_space) |
|
69
|
|
|
opt1.search(objective_function, n_iter=100, initialize={"random": 1}) |
|
70
|
|
|
|
|
71
|
|
|
assert abs(opt0.best_score - opt1.best_score) > err |
|
72
|
|
|
|
SimonBlanke /
Gradient-Free-Optimizers
