|
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
|
|
|
) |
|
17
|
|
|
|
|
18
|
|
|
data = load_iris() |
|
19
|
|
|
X, y = data.data, data.target |
|
20
|
|
|
|
|
21
|
|
|
|
|
22
|
|
|
def model(para, X_train, y_train): |
|
23
|
|
|
model = DecisionTreeClassifier(criterion=para["criterion"]) |
|
24
|
|
|
scores = cross_val_score(model, X_train, y_train, cv=2) |
|
25
|
|
|
|
|
26
|
|
|
return scores.mean() |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
def model1(para, X_train, y_train): |
|
30
|
|
|
model = DecisionTreeClassifier(max_depth=para["max_depth"]) |
|
31
|
|
|
scores = cross_val_score(model, X_train, y_train, cv=2) |
|
32
|
|
|
|
|
33
|
|
|
return scores.mean() |
|
34
|
|
|
|
|
35
|
|
|
|
|
36
|
|
|
def model2(para, X_train, y_train): |
|
37
|
|
|
model = DecisionTreeClassifier(max_depth=para["max_depth"]) |
|
38
|
|
|
scores = cross_val_score(model, X_train, y_train, cv=2) |
|
39
|
|
|
|
|
40
|
|
|
return scores.mean() |
|
41
|
|
|
|
|
42
|
|
|
|
|
43
|
|
|
search_config = {model: {"criterion": ["gini"]}} |
|
44
|
|
|
search_config1 = {model1: {"max_depth": range(2, 500)}} |
|
45
|
|
|
search_config2 = {model2: {"max_depth": range(2, 500)}} |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
def test_delete_model(): |
|
49
|
|
|
delete_model(model) |
|
50
|
|
|
|
|
51
|
|
|
opt = Hyperactive(X, y) |
|
52
|
|
|
opt.search(search_config) |
|
53
|
|
|
|
|
54
|
|
|
delete_model(model) |
|
55
|
|
|
|
|
56
|
|
|
|
|
57
|
|
|
def test_delete_model_dataset(): |
|
58
|
|
|
delete_model_dataset(model, X, y) |
|
59
|
|
|
|
|
60
|
|
|
opt = Hyperactive(X, y) |
|
61
|
|
|
opt.search(search_config) |
|
62
|
|
|
|
|
63
|
|
|
delete_model_dataset(model, X, y) |
|
64
|
|
|
|
|
65
|
|
|
|
|
66
|
|
View Code Duplication |
def test_connect_model_IDs(): |
|
|
|
|
|
|
67
|
|
|
delete_model(model1) |
|
68
|
|
|
delete_model(model2) |
|
69
|
|
|
|
|
70
|
|
|
connect_model_IDs(model1, model2) |
|
71
|
|
|
|
|
72
|
|
|
c_time = time.time() |
|
73
|
|
|
opt = Hyperactive(X, y, memory="long") |
|
74
|
|
|
opt.search(search_config1, n_iter=1000) |
|
75
|
|
|
diff_time_0 = time.time() - c_time |
|
76
|
|
|
|
|
77
|
|
|
c_time = time.time() |
|
78
|
|
|
opt = Hyperactive(X, y, memory="long") |
|
79
|
|
|
opt.search(search_config2, n_iter=1000) |
|
80
|
|
|
diff_time_1 = time.time() - c_time |
|
81
|
|
|
|
|
82
|
|
|
assert diff_time_0 / 2 > diff_time_1 |
|
83
|
|
|
|
|
84
|
|
|
|
|
85
|
|
View Code Duplication |
def test_split_model_IDs(): |
|
|
|
|
|
|
86
|
|
|
delete_model(model1) |
|
87
|
|
|
delete_model(model2) |
|
88
|
|
|
|
|
89
|
|
|
connect_model_IDs(model1, model2) |
|
90
|
|
|
|
|
91
|
|
|
split_model_IDs(model1, model2) |
|
92
|
|
|
|
|
93
|
|
|
c_time = time.time() |
|
94
|
|
|
opt = Hyperactive(X, y, memory="long") |
|
95
|
|
|
opt.search(search_config1, n_iter=1000) |
|
96
|
|
|
diff_time_0 = time.time() - c_time |
|
97
|
|
|
|
|
98
|
|
|
c_time = time.time() |
|
99
|
|
|
opt = Hyperactive(X, y, memory="long") |
|
100
|
|
|
opt.search(search_config2, n_iter=1000) |
|
101
|
|
|
diff_time_1 = time.time() - c_time |
|
102
|
|
|
|
|
103
|
|
|
assert diff_time_0 / 2 < diff_time_1 |
|
104
|
|
|
|
SimonBlanke /
Hyperactive
Loading history...