1
|
|
|
# Author: Simon Blanke |
2
|
|
|
# Email: [email protected] |
3
|
|
|
# License: MIT License |
4
|
|
|
|
5
|
|
|
import time |
6
|
|
|
|
7
|
|
|
from sklearn.datasets import load_iris |
8
|
|
|
from sklearn.model_selection import cross_val_score |
9
|
|
|
from sklearn.tree import DecisionTreeClassifier |
10
|
|
|
from hyperactive import Hyperactive |
11
|
|
|
from hyperactive.memory import ( |
12
|
|
|
delete_model, |
13
|
|
|
delete_model_dataset, |
14
|
|
|
connect_model_IDs, |
15
|
|
|
split_model_IDs, |
16
|
|
|
get_best_model, |
17
|
|
|
reset_memory, |
18
|
|
|
) |
19
|
|
|
|
20
|
|
|
data = load_iris() |
21
|
|
|
X, y = data.data, data.target |
22
|
|
|
|
23
|
|
|
|
24
|
|
|
def model(para, X_train, y_train): |
25
|
|
|
model = DecisionTreeClassifier(criterion=para["criterion"]) |
26
|
|
|
scores = cross_val_score(model, X_train, y_train, cv=2) |
27
|
|
|
|
28
|
|
|
return scores.mean() |
29
|
|
|
|
30
|
|
|
|
31
|
|
|
def model1(para, X_train, y_train): |
32
|
|
|
model = DecisionTreeClassifier(max_depth=para["max_depth"]) |
33
|
|
|
scores = cross_val_score(model, X_train, y_train, cv=2) |
34
|
|
|
|
35
|
|
|
return scores.mean() |
36
|
|
|
|
37
|
|
|
|
38
|
|
|
def model2(para, X_train, y_train): |
39
|
|
|
model = DecisionTreeClassifier(max_depth=para["max_depth"]) |
40
|
|
|
scores = cross_val_score(model, X_train, y_train, cv=2) |
41
|
|
|
|
42
|
|
|
return scores.mean() |
43
|
|
|
|
44
|
|
|
|
45
|
|
|
search_config = {model: {"criterion": ["gini"]}} |
46
|
|
|
search_config1 = {model1: {"max_depth": range(2, 500)}} |
47
|
|
|
search_config2 = {model2: {"max_depth": range(2, 500)}} |
48
|
|
|
|
49
|
|
|
|
50
|
|
|
def test_reset_memory(): |
51
|
|
|
reset_memory(force_true=True) |
52
|
|
|
|
53
|
|
|
|
54
|
|
|
def test_delete_model(): |
55
|
|
|
delete_model(model) |
56
|
|
|
|
57
|
|
|
opt = Hyperactive(X, y) |
58
|
|
|
opt.search(search_config) |
59
|
|
|
|
60
|
|
|
delete_model(model) |
61
|
|
|
|
62
|
|
|
|
63
|
|
|
def test_delete_model_dataset(): |
64
|
|
|
delete_model_dataset(model, X, y) |
65
|
|
|
|
66
|
|
|
opt = Hyperactive(X, y) |
67
|
|
|
opt.search(search_config) |
68
|
|
|
|
69
|
|
|
delete_model_dataset(model, X, y) |
70
|
|
|
|
71
|
|
|
|
72
|
|
View Code Duplication |
def test_connect_model_IDs(): |
|
|
|
|
73
|
|
|
delete_model(model1) |
74
|
|
|
delete_model(model2) |
75
|
|
|
|
76
|
|
|
connect_model_IDs(model1, model2) |
77
|
|
|
|
78
|
|
|
c_time = time.time() |
79
|
|
|
opt = Hyperactive(X, y, memory="long") |
80
|
|
|
opt.search(search_config1, n_iter=1000) |
81
|
|
|
diff_time_0 = time.time() - c_time |
82
|
|
|
|
83
|
|
|
c_time = time.time() |
84
|
|
|
opt = Hyperactive(X, y, memory="long") |
85
|
|
|
opt.search(search_config2, n_iter=1000) |
86
|
|
|
diff_time_1 = time.time() - c_time |
87
|
|
|
|
88
|
|
|
assert diff_time_0 / 2 > diff_time_1 |
89
|
|
|
|
90
|
|
|
|
91
|
|
View Code Duplication |
def test_split_model_IDs(): |
|
|
|
|
92
|
|
|
delete_model(model1) |
93
|
|
|
delete_model(model2) |
94
|
|
|
|
95
|
|
|
connect_model_IDs(model1, model2) |
96
|
|
|
|
97
|
|
|
split_model_IDs(model1, model2) |
98
|
|
|
|
99
|
|
|
c_time = time.time() |
100
|
|
|
opt = Hyperactive(X, y, memory="long") |
101
|
|
|
opt.search(search_config1, n_iter=1000) |
102
|
|
|
diff_time_0 = time.time() - c_time |
103
|
|
|
|
104
|
|
|
c_time = time.time() |
105
|
|
|
opt = Hyperactive(X, y, memory="long") |
106
|
|
|
opt.search(search_config2, n_iter=1000) |
107
|
|
|
diff_time_1 = time.time() - c_time |
108
|
|
|
|
109
|
|
|
assert diff_time_0 / 2 < diff_time_1 |
110
|
|
|
|
111
|
|
|
|
112
|
|
|
def test_get_best_model(): |
113
|
|
|
score, search_config, init_config = get_best_model(X, y) |
114
|
|
|
|
115
|
|
|
reset_memory(force_true=True) |
116
|
|
|
|